Below the Surface

Kraken Robotics is a profitable Canadian subsea-technology company^VerifiedEvidence. Profitable and growing: FY2025 revenue C$102.2M (+12% YoY) at a 24.4% Adjusted-EBITDA margin; Q1-2026 revenue +35% YoY. Subsea-technology company, HQ in Canada (Mount Pearl / St. John’s).Sources: Kraken, FY2025 results, 16 Apr 2026 · Kraken, Q1-2026 results, 28 May 2026 that has grown over a decade from a single sonar line into the full subsea stack^VerifiedEvidence. Built over a decade from one sonar line into the full stack: AquaPix interferometric/SAS sea-trials in 2013, expanding into subsea power, robotics and services.Sources: Kraken, AquaPix sonar sea-trials, 2013 · Kraken, AquaPix/MINSAS SAS · Kraken, FY2025 results, 16 Apr 2026, much of it assembled by acquisition^VerifiedEvidence. The stack was assembled by acquisition: pressure-tolerant batteries / SeaPower from the Kraken Power business (2016), plus PanGeo Subsea and the Dive Technologies lineage. Covelya is the latest, largest step.Sources: EE Power, Kraken acquires pressure-tolerant battery business, 2016 · Kraken / Dive Technologies sea-trials, 2021 · Kraken, $615M Covelya acquisition, 3 Mar 2026. It runs best-in-class 24.4% EBITDA margins^VerifiedEvidence. Audited FY2025 Adjusted-EBITDA margin 24.4%, best-in-class vs subsea/defence peers; Q1-2026 stepped up again; combined Kraken+Covelya ~24% (‘~25%’ rounds the top of the range).Sources: Kraken, FY2025 results, 16 Apr 2026 · GlobeNewswire, FY2025 results · Kraken, Q1-2026 results, 28 May 2026 on two physics-backed franchises (synthetic-aperture sonar and pressure-neutral SeaPower batteries^VerifiedEvidence. Two physics-backed franchises: AquaPix/MINSAS synthetic-aperture sonar, and pressure-neutral, housing-free Li-ion SeaPower batteries (rated to 6,000 m). Both are fielded; headline energy densities are company-stated.Sonar: synthetic-aperture (SAS)Kraken, AquaPix/MINSAS SASKraken, AquaPix sonar sea-trials, 2013Kraken, SAS at NATO REPMUS 2025Power: SeaPower batteriesKraken, SeaPower Tech BookKraken, Investor Presentation, Apr 2026Kraken, US Navy CRADA (batteries), 2021Lineage & IPEE Power, Kraken acquires pressure-tolerant battery business, 2016Pressure-tolerant battery patent (Schlumberger origin)Kraken / Dive Technologies sea-trials, 2021) and already sits inside the West’s uncrewed undersea fleet, powering Anduril’s Ghost Shark^VerifiedEvidence. Kraken is embedded in Anduril’s XL undersea fleet: SeaPower batteries power Dive-LD / Dive-XL / Ghost Shark (50–60 modules per XLUUV), MINSAS-120 sonar confirmed on the sister Dive-LD, buyer-side confirmed by Anduril’s founder. Per-hull $ is sell-side estimate; SAS on Ghost Shark is strongly indicated, not primary-confirmed.Company primary: SAS + batteries on Dive-LDKraken PR, “Dive Technologies Completes Successful Sea Trials,” 13 Apr 2021, Dive-LD ran Kraken MINSAS-120 SAS + pressure-tolerant batteries (30 missions). Dive was later bought by Anduril.Lineage, Defense One, “Anduril buys Dive Technologies,” Feb 2022.Buyer-side battery confirmationPalmer Luckey (Anduril founder), X, 12 Dec 2022, “Kraken makes batteries. We buy batteries from Kraken.”Informal first-party post; not independently re-fetched.SeaPower across the XL fleet: corroborationKraken Investor Presentation, Apr 2026, “mission-critical systems to Anduril’s underwater platforms”; “XLUUVs include 50–60 of these batteries.”Kraken SeaPower Tech Book, 2024, battery spec & per-module economics.Baird Maritime, “Vessel Review | Ghost Shark,” 4 Dec 2024, 6,000 m depth rating matches SeaPower.Naval News, “Anduril’s Dive-XL Setting New Standards,” 10 Dec 2024, COTS pressure vessels house “navigation, communications, and batteries.”Politico, VAdm Topshee (Commander RCN), 13 Feb 2026, Kraken builds the Ghost Shark batteries (“Canadian tech in that”).Kraken SAS on Ghost Shark: strongly indicated, not confirmedFor: sister Dive-LD confirmed (Kraken PR); Desjardins note (“potentially SAS”).Against: The War Zone, Dr Shane Arnott (Anduril) interview, Anduril conceals payloads; no official source names Kraken sonar on Ghost Shark.Per-hull Kraken content $: sell-side estimates onlyDesjardins via Cantech Letter, Aug 2025, ~US$2M batteries / up to ~US$8M total Kraken products per AUV.Viral “~C$10M per hull / ~C$400M-a-year” specifics trace to retail write-ups, not Kraken or Anduril, low confidence.Sonardyne (Covelya) navigation & commsDive-LD: SPRINT-Nav X, AvTrak 6, Ranger 2 / Micro-Ranger 2, Sonardyne Anduril/Dive-LD case study, 28 Nov 2024.Ghost Shark: Anduril supplier announcement (Nov 2023) names Sonardyne for sub-surface navigation; function only (Advanced Navigation also a named partner). and a second, still-unnamed U.S.-defence customer^Company claimEvidence. On its Q4 FY2025 call management said it ‘did land another large U.S. defense company’ as a battery customer, unnamed, and not independently confirmed.Source: Kraken Q4/FY2025 earnings call (transcript).

Its latest move is a step-change: the ~C$615M acquisition of the UK’s Covelya Group (closing 2 July 2026)^VerifiedEvidence. Definitive C$615M agreement to acquire Covelya signed 3 Mar 2026; all regulatory & exchange approvals received 18 Jun; close expected ~2 Jul 2026.Sources: Kraken, regulatory approval, 18 Jun 2026 · Kraken, $615M Covelya acquisition, 3 Mar 2026 · Unmanned Systems Tech, $615M deal · MassRobotics, $615M deal, turning Kraken into a larger, multi-brand subsea platform spanning sonar, power, navigation, positioning & covert comms^VerifiedEvidence. Covelya brings Sonardyne: subsea navigation (SPRINT-Nav), acoustic positioning / USBL (Ranger) and covert comms (AvTrak), adding navigation, positioning & comms to Kraken’s sonar + power: a multi-brand subsea platform.Sources: Kraken, $615M Covelya acquisition, 3 Mar 2026 · Sonardyne (Covelya), products · Sonardyne, history. Kraken is consuming a business roughly 2.5× its size^VerifiedEvidence. Covelya is large: management profile puts 2025E revenue at ~US$249–275M vs Kraken’s FY2025 ~C$102M, roughly 2.5× Kraken’s revenue base.Sources: Kraken, $615M Covelya acquisition, 3 Mar 2026 · UK Companies House (Covelya), accounts, yet bought it cheaply (~9.7× EBITDA) and financed it conservatively (~78% cash, ~0.3× net leverage)^VerifiedEvidence. Kraken bought Covelya at ~9.7–10× FY25E EBITDA vs a defence-M&A average ~14× / median ~19×, funded ~78% cash (C$402.5M receipts) at ~0.3× net debt, the most conservative of comparable transformational deals.Sources: Kraken, $615M Covelya acquisition, 3 Mar 2026 · Kraken, C$402.5M offering close, 12 Mar 2026 · Capstone, defence-M&A multiples, Dec 2025 · PCE, A&D M&A multiples · Teledyne, Valeport acquisition (precedent), 2024.

The timing rides the strongest naval and military-spending tailwind in a generation^VerifiedEvidence. Generational defence up-cycle: NATO’s 2025 Hague summit set a 5%-of-GDP path; the EU adopted ‘Readiness 2030’; SIPRI reports record world military spending.Sources: NATO, Hague Summit (5% goal), Jun 2025 · European Commission, Defence Readiness 2030 · SIPRI, world military expenditure 2025, with demand signals visible in the real world (a naval disruption around the Strait of Hormuz^VerifiedEvidence. What I first logged as a weak social-media signal is now a primary-documented naval crisis. Iran closed the Strait from late February 2026; a US counter-blockade followed from mid-April; Gulf crude shipments fell ~95% and LNG ~99% at the trough, on a lane carrying ~20% of the world’s oil and LNG. CENTCOM began clearing mines on 11 April 2026, and Iran reportedly cannot locate its own minefields, which is precisely the problem sonar-led MCM exists to solve.PrimaryUS CENTCOM, mine-clearance mission begins in the Strait of Hormuz (11 Apr 2026)UK House of Commons Library, briefing CBP-10636, the 2026 Hormuz crisisCorroborationMilitarnyi, Iran cannot locate its Hormuz mines (citing NYT, 11 Apr 2026)Jerusalem Post, same NYT-sourced reporting, independently carriedStrongest contraryAl Jazeera, how the Hormuz minesweeping works (25 Jun 2026): the reported US–Iran framework puts clearing obligations on IranCrisis demand can de-escalate as fast as it arrived; my thesis prices the structural cycle, not this event., rising China–Taiwan tensions^VerifiedEvidence. Indo-Pacific friction is rising, e.g. new US Taiwan legislation (Dec 2025) and elevated cross-strait tension, supporting allied undersea / mine-countermeasures demand.Source: Reuters, new US Taiwan legislation, Dec 2025, and the defence of UK and European undersea infrastructure against Russia^VerifiedEvidence. Protecting undersea infrastructure is now explicit allied policy: NATO ‘Baltic Sentry’ after Baltic cable incidents, the UK’s ‘Atlantic Bastion’, and the EU/Canada SAFE pact.Sources: Royal Navy, ‘Atlantic Bastion’, Dec 2025 · EU Council, SAFE defence pact (Canada), Dec 2025 · CNN, AUKUS seabed-cable defence, May 2026) in a world underprepared to protect its critical chokepoints^My viewEvidence. My view, backed by oversight reporting: allied mine-countermeasures / seabed-warfare capacity lags the threat: the US GAO flags the Navy’s XLUUV undersea-minelayer program years behind schedule and over budget.Source: USNI / GAO, Navy XLUUV minelayer behind schedule.

Much of the mispricing is mechanical^My viewEvidence. My view: much of the mispricing is mechanical: venture / OTC listing, index & screen exclusion, and trough-year optics rather than fundamentals.Source: my assessment (see Valuation section).. A TSX-Venture and OTC listing keeps a ~C$2B company off most institutional radars^Verified · factsEvidence. Listed on TSX-Venture (PNG) and OTCQB (KRKNF); at ~C$6.34 the market cap is ~C$1.95B. A venture / OTC listing keeps it off many institutional screens; the ‘off-radar’ point is my inference.Sources: StockAnalysis / TMX, PNG market data (delayed) · Google Finance, PNG:CVE (30 Jun 2026 close) · TMX, TSXV / listing, and simple screens call it ‘priced to perfection’ on a trough-year P/E north of 650×^My calcEvidence. On trough-year FY2025 earnings a simple P/E screens north of 650×, an artefact of a depressed denominator and non-cash / acquisition items, not operating reality.Source: my calc from Kraken, FY2025 results, 16 Apr 2026 · StockAnalysis / TMX, PNG market data (delayed) · Google Finance, PNG:CVE (30 Jun 2026 close)., while missing the more relevant ~23× EV/EBITDA lens and the structural change Covelya brings^EstimateEvidence. On forward EBITDA the more relevant lens is ~23× EV/EBITDA at the 30-Jun close of C$6.34 (FY26E combined basis), broadly in line with subsea / defence peers (Teledyne ~20×, Kongsberg ~29–31×).Sources: Capstone, defence-M&A multiples, Dec 2025 · StockAnalysis / TMX, PNG market data (delayed). At ~C$6.34 (30 June 2026 close)^Single source · as of Jun 2026Evidence. At the 30 Jun 2026 close, TSXV:PNG C$6.34 (+1.6% on the day; OTCQB:KRKNF; mkt cap ~C$1.95B, 307.2M shares out). Delayed market data, time-sensitive and will drift.Source: StockAnalysis / TMX, PNG market data (delayed), the market is pricing the wrong company.

The mechanics have receipts. There are two ‘Krakens’ in naval autonomy, and the national press has already attributed the other one’s Royal Navy win to this company^VerifiedEvidence. Two distinct companies share the name: Kraken Technology Group (Fareham, UK) builds uncrewed surface vessels, is itself an Anduril partner, and won the Royal Navy’s £12.3M, 20-boat USV buy in March 2026. The i Paper has already run the confusion in print, folding that award into a story about this company’s minehunting kit, a misattribution I logged and corrected in my base. Kraken Robotics builds no USVs.Sources: Anduril, partnership with Kraken Technology Group (the other Kraken) · Royal Navy, 20-USV procurement award, 11 Mar 2026 · The i Paper, the article that mixes the two, Jun 2026. The spring-2026 rotation drained the whole defence cohort while memory, photonics and the SpaceX listing absorbed the market’s attention, Kraken fell with its sector, not out of it^Verified · my registerEvidence. From my macro-context sweep (Jun 2026): the US A&D ETF sat roughly flat on the year and ~12% below its early-March peak while CNBC ran “Defense stocks have floundered” (28 Apr 2026, named, the exact link is truncated in my register, so no URL) and Bernstein called it rotation, not fundamentals. Where the capital went is documented: three DRAM makers crossed ~US$1T market caps within weeks, Micron ran ~+800% in a year, and SpaceX priced the largest IPO in history on 12 Jun (~US$75B raised, ~$1.77T at pricing). Kraken fell with its cohort, not out of it. Caveat: rotations reverse without notice; figures are mid-Jun snapshots.Sources: CNBC, TrendForce, Nasdaq/Fortune coverage, all resolved in my register · StockAnalysis, PNG.V price history (dated), and the venue drags on top: the Venture composite spent the year down while the senior board rose^Verified · in partEvidence. At the 12-Jun-2026 snapshot the S&P/TSX Venture Composite was −2.9% YTD against +8.7% for the senior-board Composite, from a study I commissioned on senior-board-versus-venture behaviour (public price-return series, TMX and S&P index methodology, plus the academic literature). Its own caveat, at full weight: the senior-board small-cap index was also violent, so much of the swing is the small-cap factor, not the venue, graduation buys a moderate benefit, not immunity.Sources: My commissioned TSX-vs-TSXV study (methodology as stated; research file, no public URL). Then the optics: screens that auto-reject a trough-year P/E north of 650× re-engage after the close, when the combined entity prints on sane multiples^My calcEvidence. Trailing P/E ran ~650–760× depending on the price snapshot (my arithmetic off the FY2025 trough net income; ~630–670× at the 30-Jun C$6.34 close; ~757× at C$7.04, 12 Jun) with ~21× sales and ~83× EV/EBITDA standalone, numbers a crude screen rejects unread. My model of the combined entity prints roughly ~62× forward P/E, ~23× EV/EBITDA and ~5.2× P/S at the 30-Jun close (mid-June snapshot: ~70×/~24×/~5.6×), ranges screens and buyers actually engage with. The mechanism is standard M&A finance: completion removes the overhang and lets the market re-rate on pro-forma numbers. Bound, honestly: systematic factor-selling bites hardest on index-held large caps, which venture-listed Kraken is not, this mechanism is real but bounded.Sources: BIS Quarterly Review, de-grossing / factor selling in volatility · Yahoo Finance, deal close removes acquirer overhang (worked example) · StockAnalysis, PNG.V (basis, dated). Add a register that has largely turned over since the twelvefold climb, newcomers see only a vertical chart they assume they missed, and you get a stock most institutions cannot hold, most screens reject, and most readers misread. That is not efficiency; that is the opportunity.

On the model, the market is still pricing Kraken below the fair-value base case^My modelEvidence. On my EV/EBITDA scenario model, ~C$6.34 sits below the base-case exit, a model output, not market consensus; the sum-of-the-parts view tempers it.Source: my valuation model, inputs: Kraken, FY2025 results, 16 Apr 2026 · UK Companies House (Covelya), accounts., and the setup is favourably skewed^My viewEvidence. My judgment, not a sourced fact: the asymmetry is the output of the bull / bear weighing set out across this thesis.Source: my assessment (see Risks & Valuation sections).: with organic growth seemingly re-accelerating^VerifiedEvidence. Growth is re-accelerating: Q1-2026 revenue +35% YoY; FY2025 record order intake; $35M SeaPower battery sales announced Jan 2026; SAS deployed at NATO REPMUS 2025.Sources: Kraken, Q1-2026 results, 28 May 2026 · Kraken, FY2025 results, 16 Apr 2026 · Kraken, $35M SeaPower battery sales, Jan 2026 · Kraken, SAS at NATO REPMUS 2025, there is a credible path to the bullish end of a polarised Street (~C$11–14, against a ~C$6.50–14 range)^VerifiedEvidence. Coverage is polarised: Desjardins Buy, target C$14 (Street high); ATB Capital Markets (N. Boychuk, ex-Cormark) Sell C$6.50; ~C$10.80 consensus across ~7 dealers (Cormark, Canaccord, Desjardins, National Bank, ATB, Ventum, Eight Capital).Sources: Desjardins via Cantech Letter, target C$14, Aug 2025 · Financial Post (Yahoo), coverage / consensus if Covelya integrates cleanly^My view · riskEvidence. My view: integration is the key swing variable: cross-border (UK / US / Canada) execution and CFIUS / ITAR are the main risks; see the bear case.Source: my assessment (see Risks & bear case).. The proof points are concrete and near-term: the first combined reporting cycle, backlog conversion, margin preservation, and cross-sell evidence.

Few are paying attention, and even fewer have analysed it to the depth that follows.

Kraken’s portfolio grew the way the company did, one capability at a time, usually bought, then scaled. Walk the block diagram of a modern uncrewed underwater vehicle and almost every box, perception, power, propulsion, imaging, and (with Covelya) navigation, positioning and communications, is now something Kraken makes^My viewEvidence. My synthesis of the logged product register: across perception, power, propulsion, imaging, and with Covelya, navigation, positioning & comms, Kraken makes nearly every subsystem of a modern UUV.Source: my assessment of the product register (see cards below for the underlying primaries).. Specs below are company-stated unless independently corroborated.

The provenance is worth thirty seconds, because it is routinely told wrong. The AquaPix IP came out of Marport by negotiated spin-out in January 2013, not a receivership fire-sale, and within two years Canada’s DRDC had flown it on the Franklin Expedition search that found HMS Erebus^Verified · my researchEvidence. From my four-agent primary-source pass over the 2013–2015 releases and federal registry records (de-mythologising check included): the AquaPix IP left Marport by negotiated spin-out announced 7 Jan 2013, not the receivership fire-sale of market lore (Marport’s fishing division went separately to Airmar in Oct 2013), and Canada’s DRDC deployed Kraken SAS on the 2014 Franklin Expedition search that found HMS Erebus. The 2012 incorporation date rests on a single registry source, flagged as such.Sources: Kraken/GlobeNewswire 2013–2015 releases + federal corporate registry (resolved in my register; no single public URL). The company reached the market in February 2015 not by IPO but through a capital-pool shell, at roughly a C$10M valuation on C$1.9M of revenue, the entire firm was then priced at less than a third of the single Danish minehunting package it would win five years later^Verified · my readEvidence. A Capital Pool Company qualifying transaction into the shell Anergy Capital Inc. (TSXV: ACA.H): 2.25-to-1 consolidation, renamed Kraken Sonar Inc., closed 18 Feb 2015, roughly 71.0M shares at a $0.15 deemed price, ~C$10M nominal market cap, against ~C$1.9M of FY2015 revenue. The comparison is my arithmetic: the Royal Danish Navy’s single C$36M KATFISH package of 2020 was worth more than three of the whole 2015 company.Sources: Kraken CPC closing + ‘starts trading’ releases, Feb 2015 (my register) · StockAnalysis, PNG.V full price history.

The founding line is AquaPix itself, the interferometric synthetic-aperture sonar, miniaturised into the MINSAS family, 60 / 120 / 180 / 240, named for array length, compact enough for small-diameter AUVs. The claim that matters: constant ~3.3×3 cm resolution regardless of range^Company claimEvidence. Company-stated specs: MINSAS 60/120/180/240 constant-resolution SAS, ~3.3×3.0 cm real-time / ~2×2 cm post-processed, independent of range; 100–200 m swath per side. AquaPix IP acquired 2013; miniaturised from 2014. Not independently bench-tested.Sources: Kraken, AquaPix/MINSAS product page & flyer (V.050125) · Kraken, AquaPix sea-trials, 2013, where conventional side-scan blurs with distance. The business model matters more: this is a merchant sonar that ships inside other people’s vehicles, Anduril’s Dive-LD, Teledyne’s Gavia and SeaRaptor, the standard fit on Huntington Ingalls’ REMUS 620^VerifiedEvidence. The merchant fits are order-verified: MINSAS-120 ran on Anduril’s Dive-LD in 2021 sea trials; HII ordered MINSAS for REMUS 620 (US$1M, Nov 2023) where it is the flagship sonar; Teledyne (Gavia / SeaRaptor) sits inside the Dec-2025 $12M order; a record number of navies deployed Kraken SAS at NATO REPMUS 2025.Platform wins (company-primary)Kraken / Dive Technologies sea-trials (MINSAS-120 + batteries), Apr 2021Kraken, MINSAS order from HII for REMUS 620, Nov 2023Kraken, $12M SAS + battery orders (incl. Teledyne), Dec 2025Independent / buyer-side corroborationHII, REMUS 620 torpedo-tube launch & recovery, 2025Kraken, SAS at NATO REPMUS 2025. The towed line, KATFISH, grew from a 2016 Elbit order into the SeaScout minehunting package the Danish and Polish navies bought^VerifiedEvidence. KATFISH grew from Elbit’s 2016 order (>C$15M) into the SeaScout package, actively-stabilised dual-frequency SAS towfish + Tentacle winch + ALARS autonomous launch-and-recovery, bought by the Royal Danish Navy (~C$36M, 2020, with a 17-year support tail) and the Polish Navy (Kormoran II MCMVs).Sources: Kraken, $36M Danish Navy contract, 8 Sep 2020 · Kraken, Polish Navy minehunting systems, 1 Sep 2020.

SeaPower, built in Rostock, is the highest-margin, fastest-growing product, and the architecture is the differentiator: pouch cells potted in silicone, no oil, no pressure housing, pressure-neutral to 6,000 m. Deleting the titanium case pays off more the deeper you go: ~260 Wh/L and ~3.2 kg/kWh in-water weight lead the depth-rated field^Company claimEvidence. Architecture verified by fielding; headline numbers are company-stated: Li-ion NMC pouch cells potted in silicone polymer, no oil, no pressure housing, pressure-neutral to 6,000 m / 660 bar; ~260 Wh/L, ~145 Wh/kg system-level, >2,000 cycles to 80%; ~2 kWh modules chaining toward multi-MWh banks (flyer range: 2 kWh–5 MWh).Sources: Kraken, SeaPower product page · Kraken, SeaPower flyer (V.050225) · Kraken, SeaPower Tech Book. And the battery has a life beyond vehicles: the Verlume collaboration points SeaPower at seabed energy storage and resident-AUV charging^VerifiedEvidence. Strategic collaboration announced 27 Aug 2025: SeaPower paired with Verlume’s seabed energy-storage and resident-AUV charging infrastructure. Partnership verified; revenue contribution still speculative, treat as optionality, not forecast.Source: Verlume × Kraken strategic collaboration, 27 Aug 2025.

The same Rostock plant builds a second product from the same playbook: SeaThrust, a rim-driven, shaftless thruster, no gearbox, no central shaft, seawater-lubricated and rated to the same 6,000 m^Company claimEvidence. SeaThrust exists as a shipping product (verified); specs are company-stated: pressure-tolerant rim-driven thruster, seawater-lubricated, no gearbox or central shaft, rated 6,000 m, built by the same Rostock (Kraken Power GmbH) unit as SeaPower.Source: Kraken, product one-pagers (SeaThrust / SeaPower / KATFISH). Pair it with the battery and Kraken is one vendor for both the quiet power source and the quiet propulsor.

The capacity story in management’s own sequence^Their words · call transcriptEvidence. Q4-FY2024 call (28 Apr 2025): Rostock “running full out”; 2025 capacity largely secured; second XL-UUV customer targeted by year-end. Q4-FY2025 call (16 Apr 2026): “did land another large U.S. defense company”; battery revenue +~30 percent YoY; higher-energy-density packs offsetting price pressure at better $/kWh. Company claims on recorded calls; the second customer remains unnamed, not independently confirmed.Sources: Q4/FY2024 call (transcript) · Q4/FY2025 call (transcript): April 2025, the German plant “running full out”, targeting a second XL-UUV battery customer by year-end; April 2026, that second large US defence customer landed (unnamed), battery revenue up roughly 30 percent, and price pressure answered with higher-energy-density packs at better $/kWh. Said on the record, a year apart, in the order the thesis needs.

And the absence everyone eventually asks about: SeaPower is deliberately un-patented^My read · registry-groundedEvidence. The factual layer is verified by my patent-registry pass (six parallel searches over Google Patents/Justia/WIPO legal-event records, Jun 2026): no granted Kraken-owned pressure-tolerant-battery patent exists, and the closest pressure-neutral family belongs to a third party (Alfred-Wegener-Institut). The strategic reading is mine: a patent is a publication, it teaches the formulation and process to every competitor, including those already holding adjacent pressure-neutral IP, in exchange for twenty years of protection Kraken would have to litigate to enforce. A potted, silicone-encapsulated pack is genuinely hard to reverse-engineer from a fielded unit; the chemistry, compensation and potting process live in the Rostock and Dartmouth lines and in heads. Filing would sharpen rivals today for an edge that expires tomorrow. Kraken has never publicly stated this rationale, the choice is visible in the register, the reasoning is my inference.Sources: my registry pass (patent-office legal-event records, in my evidence base) · Google Patents, assignee records. That is not an oversight, it is the classic trade-secret trade-off. A patent publishes the recipe: the cell selection, the silicone potting chemistry, the pressure-compensation process, handed in full to every rival, including the third parties who already hold adjacent pressure-neutral IP, in exchange for a twenty-year clock and the burden of policing it. Kept secret, the edge lives in the manufacturing lines and the people, is brutally hard to extract from a fielded pack, and never expires. The cost of that choice is honest and carried in my bear case: trade secrets protect nothing in law if a rival independently gets there, which is exactly why the moat section grades this edge on physics and know-how, not paper.

PanGeo (2021) added 3D acoustic imaging beneath the seabed, cable routing, UXO, offshore-wind geotechnics, and turned a product company into a recurring-services one; 3D at Depth (2025) added millimetre-grade subsea LiDAR^VerifiedEvidence. PanGeo Subsea (acquired 2021) added the Sub-Bottom Imager / Acoustic Corer 3D acoustic-imaging services line; 3D at Depth (2025) added the SL-series subsea LiDAR (mm-level metrology). Kraken’s own ThunderFish AUVs remain niche / IRAP-stage, subsystems are the centre of gravity.Sources: Kraken, products overview · Kraken, FY2025 results (services segment), 16 Apr 2026. Kraken builds its own AUVs (ThunderFish) only at the margin: the centre of gravity is subsystems, picks and shovels, whichever platform wins.

The gaps left in the stack were navigation, positioning and covert communications, precisely what Covelya closes: Sonardyne’s reference-standard GPS-denied suite, Wavefront’s defence sonars, EIVA’s survey software, Voyis optics, Chelsea sensing^VerifiedEvidence. Covelya closes the remaining layers: Sonardyne’s GPS-denied suite (SPRINT-Nav inertial navigation, Ranger 2 USBL positioning, AvTrak transponders, Modem 6 acoustic + BlueComm optical comms), Wavefront defence sonar (Solstice, Sentinel, Vigilant), EIVA’s NaviSuite software, Voyis optical inspection, Chelsea ocean sensing.Sources: Sonardyne (Covelya), products · Sonardyne, product knowledge base · Chelsea Technologies, datasheets. Two overlaps are real, Solstice against Kraken’s SAS, Voyis against 3D at Depth, and will need rationalising; the bear case owns that question^My viewEvidence. My read: two internal overlaps will need rationalising post-close, Wavefront’s Solstice against Kraken’s own SAS, and Voyis imaging against 3D at Depth (plus Kraken’s in-house SeaVision line). A product-rationalisation question, carried to the bear case.Source: my assessment from the two product registers (see Bear case)..

On sonar, the edge is architectural and commercial, not raw physics: SWaP-C, open vehicle-agnostic integration, COTS pricing, and the battery bundle. Kongsberg’s HISAS and Thales’s SAMDIS are peer-or-better on pure resolution, and Kongsberg opened HISAS to other platforms in February 2026, so the contest has moved to integration and software, ground Kraken chose early^My viewEvidence. My competitive deep-research synthesis, a multi-agent pass across ~100 cited manufacturer specs, procurement records and defence-press sources, compiled into my evidence base; no single public link: Kraken’s SAS edge rests on SWaP-C, open vehicle-agnostic architecture, COTS pricing, real-time GPU processing and the SeaPower bundle, not superior raw acoustics; Kongsberg HISAS and Thales SAMDIS are peer-or-better on resolution. Strongest contrary datapoint: Kongsberg opened HISAS to non-HUGIN platforms in Feb 2026 (logged in the project competitive file), eroding the ‘only open SAS’ line.Sources: my competitive-landscape deep-research file, a multi-agent synthesis, ~100 footnoted citations (underlying specs are vendor figures, so graded soft) · Kraken, SAS product page. On batteries the moat is more physical and more durable, but it carries no public S9310 naval-safety certification^Verified · both waysEvidence. The battery leads the depth-rated field on volumetric density (~260 Wh/L) and in-water weight (~3.2 kg/kWh), physics that improves with depth. The honest counterweight: no public US-Navy S9310 safety certification. Kraken’s own language is ‘designed for compliance with’ Navy standards; competitor GD-Bluefin has held formal Navy safety approval since 2013.Spec lead (company-stated)Kraken, SeaPower page (‘compliant with US Navy standards’)Kraken, SeaPower flyer (‘designed for compliance’)Kraken, US Navy CRADA on pressure-tolerant batteriesStrongest contrary sourceGD Mission Systems, Bluefin subsea Li battery receives Navy safety approval, 2 Apr 2013, and the estate paradox is that the flagship is essentially un-patented in Kraken’s own name: SeaPower’s defensibility is trade-secret and manufacturing know-how, harder to reverse-engineer, never time-protected^Verified · registryEvidence. Patent-office legal-event records (my registry pass, Jun 2026, six parallel searches over Google Patents, Justia and WIPO assignment/legal-event records): the estate runs to ~88 records, deep in sub-bottom imaging and subsea LiDAR (live continuations granted into 2026), but thin exactly at the flagships. The Marport-lineage SAS patents went to competitor Airmar after receivership; no granted Kraken-owned pressure-tolerant-battery patent exists, and the closest pressure-neutral family belongs to Germany’s Alfred-Wegener-Institut. SeaPower’s moat is trade-secret and manufacturing know-how, harder to copy, never time-protected. ~44 records sit Covelya-side and transfer only on completion.Registry sources (primary legal-event data)Google Patents, assignee / legal-event recordsJustia, patents assigned to Sonardyne InternationalCounts are floors, not ceilings: some office file-wrappers were machine-blocked to the research pass; ownership read from registry fields, not a legal title opinion..

The substitution threats, hydrogen fuel-cells at the ultra-endurance tier, resident seabed docking, are real, rising, and mission-bounded: fuel-cell hybrids still need a Li-ion buffer pack (content Kraken sells), and the core hours-to-days survey and MCM market stays lithium-ion^Verified · boundedEvidence. Substitution is real but mission-bounded. Ultra-endurance goes hydrogen: the US Navy / DIU CAMP effort selected Cellula’s fuel-cell Guardian (45+ days, ~5,000 km claimed), yet every fuel-cell AUV still carries a pressure-tolerant Li-ion buffer pack, content Kraken sells. Resident docking attacks the need to carry energy: Equinor’s Njord resident drone ran 165 days with 280 dockings on an inductive seabed charger. Core survey / inspection / MCM missions (hours to ~2 days) remain squarely Li-ion; the physics framing is from technical studies I commissioned (user-supplied).Fuel-cell frontierDefense Post, Cellula Guardian selected under US Navy/DIU CAMP, Apr 2026UST, Cellula Envoy >2,000 km on hydrogen, Apr 2026Resident / docking frontierOE Digital, resident-AUV landscape (Njord 165 days / 280 dockings)Verlume × Kraken (supply-side hedge), Aug 2025Endurance/range figures are vendor claims; the batteries-vs-fuel-cell physics synthesis is a study I commissioned, not independently peer-reviewed..

The moat verdicts above are assertions until you see the table, so here are both tables. Batteries first^Verified table · vendor specs beneathEvidence. My battery-comparison register: every row compiled from the vendor’s own published spec sheet or programme record (so individual cells inherit vendor-claim reliability); the comparison across rows is mine. The verdict it supports: SeaPower leads the depth-rated field on volumetric density and in-water weight because the pressure-neutral architecture deletes the titanium housing, the deeper you go, the more that choice pays. The honest columns: gravimetric density merely ties surface-grade Li-ion; SeaPower’s peak current is not published (its continuous ratings are modest); and the clearest public naval-safety certificate in the table belongs to a competitor.Key public recordsKraken, SeaPower Tech Book (specs) · GD Bluefin, US Navy safety approval, 2013 · Kraken, NSWC Carderock CRADAFuel-cell rows are a different axis (system endurance, not pack density), the substitution question is handled in the moat read above., read the architecture column and the in-water column together, because that is the whole argument:

And sonar^My matrix · red-teamedEvidence. Distilled from my SAS deep-research matrix, a multi-agent synthesis with ~100 footnoted citations to manufacturer specs, procurement records and defence press. Competitor cells are vendor specifications (soft grade); contracts and platform fits are separately verified. I red-teamed my own report before using it: its “captive to HUGIN” label was overtaken by Kongsberg opening HISAS in Feb-2026 (corrected below), and I down-rated its “durable” edge wording to medium-high. Rival unit prices (HISAS/SAMDIS/UMISAS) are quarantined, they live inside classified or bundled procurements and I will not invent them; Kraken’s own COTS prices are dated 2013 and shown as such.Verified anchors inside the matrixDanish/Polish/RCN contracts and the HII REMUS-620 fit are order-verified (receipts in the scoreboard below); Thales SAMDIS now also on REMUS 620 (DSEI-2025), the erosion flag carried honestly., where the read is different: the top tier has largely solved the physics, so competition has moved to architecture, integration and price, which is the ground Kraken picked:

What the two tables say together: in power, the lead is physical and grows with depth, peak current and the missing S9310 certificate are the honest costs. In sonar, nobody wins on resolution any more; Kraken wins on being the sonar anyone’s vehicle can carry, at a price that shows up in COTS catalogues rather than classified bundles, and the one “rival” below true SAS is now in-house.

I ran the whole estate through the patent offices rather than taking anyone’s word for it^Verified · my registry passesEvidence. Three passes over the primary patent-office record built this picture: a full assignee/inventor census (88 records: ~54 granted, ~24 pending, 8 PCT, plus one US design patent, across eleven jurisdictions, US 45, UK 13, WIPO/PCT 8, Canada 7, Europe 5, tail to Brazil/Australia/Germany/NZ/China); a chain-of-title legal-event pass confirming the assignments are office-registered, not merely asserted (3D at Depth renamed to Kraken Robotics Services US on 7 Jan 2026; PanGeo to Kraken Robotics Services Ltd, 24 Jan 2023); and a prosecution-timeline pass over the pending families.Registry sources (primary legal-event data)Google Patents, assignee, family and legal-event records · Justia, patents assigned to Sonardyne InternationalCounts are floors, not ceilings: some office file-wrappers were machine-blocked to my research pass and German utility models were not fully enumerable. “Owned” reflects registry fields cross-checked against legal events, an investment-grade read, not a freedom-to-operate opinion.:

The depth is real where it matters, and honest where it doesn’t. Two clusters are genuinely deep: sub-bottom imaging (the PanGeo lineage, ~13 records with granted US/EP/CA/AU members, e.g. US9030914B2 and US11960044B2) and subsea LiDAR, which is not just broad but actively prosecuted^Verified · live prosecutionEvidence. The LiDAR family is a live continuation chain: US10502829B2 (2019) → US11125875B2 → US12146951B2 → US12585019B2, granted 24 March 2026, three months ago, with Australian (AU2023419203) and Canadian members still in examination. The pending depth-perception family reads onto shipping product: Voyis launched “Discovery Stereo Perception” on 9 March 2026. An estate spawning granted continuations in 2025–26 and feeding launches is a living asset, not a filing-cabinet relic.Source: my prosecution-timeline pass (Google Patents legal-event tables + IP Australia AusPat, Jun 2026)., the newest grant landed in March 2026 and pending members still feed shipping product. The chain of title behind both clusters is registered at the offices themselves, not merely claimed.

The honest side of the ledger: the SAS cluster is thin, the founding sonar patents from the Marport lineage went to a competitor after receivership, so the AquaPix franchise rests on the purchased IP and accumulated know-how, which is exactly why I grade the sonar moat architectural rather than legal; the flagship battery is un-patented by choice (the trade-secret trade-off, above); and the most valuable defence IP in the combined group is Covelya’s, transferring only at completion^Verified · contingent on closeEvidence. Roughly 44 of the 88 records sit on the Covelya side and transfer only on completion, including the granted Wavefront/Sonardyne passive-detection families behind the Sentinel IDS counter-diver/counter-UUV system (GB2606160B and GB2606159B, both granted 2024, US and EP members pending) and the positioning, acoustic-comms and time-synchronisation families at Sonardyne’s core (e.g. US12498391B2, US9939527B2, US11015995B2). Until close, the strongest defence IP in the combined estate is, legally, still someone else’s, one more reason completion is the hinge event.Sources: Justia, Sonardyne assignee record · my registry passes (evidence base).. Read together: the paper moat is deep in imaging and optics, deliberate in power, thin in sonar, and the estate is growing, not ageing.

Moat claims are cheap; order flow is not. This is the platform-level record^Verified · my registerEvidence. My platform-level win/loss register, compiled from primary order releases, OEM brochures and trade press, one row per platform, each row individually sourced in my evidence base. Tally: 8 confirmed platform wins (two of them inside rival primes’ own vehicles), 1 outright RFP loss, 3 platforms where rivals hold the ground, 1 erosion flag.Win receipts (company / buyer primaries)Kraken, Dive-LD (SeaPower + MINSAS-120), 2021 · Kraken, HII REMUS 620 order, Nov 2023Kraken, Danish Navy ~C$36M, Sep 2020 · Kraken, Polish Navy Kormoran II · Kraken, $12M orders incl. Teledyne, Dec 2025EUROATLAS GreyShark brochure, “SAS courtesy KRAKEN Robotics”, 2025 · Naval News, GreyShark × Rheinmetall (150/yr → 500)The other side of the ledgerSingapore Navy MCM RFP (~C$50M) lost to Thales, single-source in my register, held provisional. BAE’s Herne XLUUV carries a Thales sonar mast; Kongsberg keeps HUGIN captive to its own HISAS; and HII, a Kraken customer, now also integrates Thales SAMDIS on REMUS: the merchant moat is real but contested., and its shape is the merchant thesis in miniature: Kraken content ships inside rivals’ vehicles (HII, Teledyne, EUROATLAS) while the two losses of ground are exactly where my bear case says they are, prime bundles and captive stacks.

Positioning, honestly: weaker overall than Kongsberg’s full captive stack; ~parity with Teledyne and Exail in subsystems (Teledyne being both rival and customer); weaker than Thales on prime MCM bundles, which is precisely why the merchant, vehicle-agnostic lane is the right one to own.

On 3 March 2026 Kraken signed to acquire the UK’s Covelya Group for ~C$615M, and unusually for a story this retail-driven, the terms are readable in the executed Share Purchase Agreement, not just the press release: C$480M in cash plus C$135M satisfied by 15,882,352 Kraken shares at C$8.50^Verified · primary contractEvidence. Consideration legs read from the executed SPA (3 Mar 2026): Base Cash CAD$480,000,000 + CAD$135,000,000 in 15,882,352 Consideration Shares at CAD$8.50. The cash leg is a GBP-denominated locked-box with a true-up (cash cap, debt deduction, working-capital collar, CAD$9M holdback), so the realised outflow floats with GBP/CAD, “C$615M” is a signing-date sum, not a guaranteed settlement.PrimaryCovelya Share Purchase Agreement, 3 Mar 2026, SEDAR+ filing 06406273 (access via sedarplus.ca; the filing ID is the canonical reference, deep links are blocked).Kraken, $615M Covelya acquisition release, 3 Mar 2026CorroborationEnergy Voice, deal coverage (≈£335M)C$615M ≈ £335M ≈ US$450M; the US$ figure is FX-derived, not a deal term. Some true-up inputs are redacted in the filing.. The financing closed before the deal did: a C$402.5M subscription-receipt offering (12 March, over-allotment fully exercised) sits in escrow, a C$150M committed term facility sits undrawn, and if the deal had failed by 31 December 2026 receipt-holders would simply have been refunded with interest^VerifiedEvidence. 47,353,550 receipts at $8.50 = ~$402.5M gross, closed 12 Mar 2026, escrowed with Computershare; C$150M five-year term facility (Bank of Nova Scotia) committed and undrawn; refund-with-interest backstop to a 31 Dec 2026 long-stop. With approvals received, the backstop scenarios fall away.Sources: Kraken, C$402.5M offering close, 12 Mar 2026 · Kraken, deal release (facility terms), 3 Mar 2026. The gauntlet of approvals that made this the thesis’s largest binary risk, UK national-security review, US CFIUS, Danish FDI, ITAR, TSXV, cleared in one omnibus announcement on 18 June 2026, with closing expected 2 July^VerifiedEvidence. All regulatory and stock-exchange approvals received 18 Jun 2026; close expected 2 Jul 2026 subject to customary conditions, on schedule vs the “near end of Q2” guidance. The SPA’s conditions stack was exactly these four FDI clearances plus TSXV.Sources: Kraken, regulatory approval, 18 Jun 2026 · SPA conditions: SEDAR+ filing 06406273.

A cross-border defence acquisition has to survive every security regulator it touches, and this one touched five^Verified · my deal trackerEvidence. My clearance-by-clearance tracker, built from the SPA’s conditions, the statutory regimes engaged, and the public record, resolved in full by Kraken’s omnibus announcement of 18 June 2026: all regulatory and stock-exchange approvals received, close expected on or about 2 July. The hard gates: TSXV; UK NSIA mandatory clearance (the central risk, Covelya sits squarely in the Act’s Defence and Dual-Use mandatory sectors); US CFIUS (Sonardyne’s Houston defence business passing to foreign control); Denmark’s Investment Screening Act (EIVA); and the ITAR/DDTC 60-day change-of-control notice, filed 5 February. Cleared without a single public call-in, phase-2 referral or remedy.PrimaryKraken, omnibus regulatory approval, 18 Jun 2026 · Share Purchase Agreement conditions (SEDAR+ filing, in my evidence base)The honest residualsThe UK CMA’s voluntary merger regime leaves a four-month own-initiative window after completion (no case on the register, product overlap is narrow); and government-contract novations on change of control are routine but administratively heavy. Neither was a closing condition. Carried, not hidden.. On 18 June, one omnibus release resolved the lot, every gate, cleared, three and a half months after signing, which is why the merger-risk discount still sitting in the price reads to me as a market pricing a risk that has already expired:

Read what the clean run means beyond timing: five security establishments, including the UK’s, under a mandatory defence-sector regime, examined a foreign takeover of their subsea-defence supplier and let it pass without a remedy. That is external validation of the combination’s acceptability that no amount of investor-deck language could buy.

Covelya is not a startup, it is a fifty-year-old anchor with four bolt-ons. Sonardyne (founded 1971, Hampshire) is the reference name in subsea positioning, navigation and acoustic communications and still makes up roughly 64% of group revenue; around it sit Wavefront (defence sonar), Chelsea (ocean sensing), EIVA (survey software, Denmark), Voyis (optical imaging, Canada) and Forcys, the defence brand that fronts them^Verified · registriesEvidence. Group shape and financials read from statutory filings, not the deal deck: turnover £70.4M → £101.0M (+43%) → ~£122.3M across FY2022–24 (~C$214M), net assets £86.3M, cash £31M, 617 employees, two 2024 King’s Awards; Sonardyne International standalone £78.7M (~64% of group). Seller = family-controlled Sonardyne Holdings Ltd; never private-equity-backed; its single Lloyds charge was cleared 23 Feb 2026, days before signing.Primary registriesUK Companies House, Covelya Group Ltd (12493148), statutory accountsSonardyne, corporate history (1971–)Conversion note£101.0M (FY2023) ≈ C$172M and ~£122.3M (FY2024) ≈ C$214M at the rates used in the model; currency is stated per figure because the headline deal is Canadian dollars.. The group was assembled the same way Kraken was, operator-led, one best-of-breed piece at a time: Wavefront in 2008, Chelsea in 2018, EIVA in 2019, 2G Robotics (now Voyis) in 2020, Forcys built internally in 2022. Kraken is buying a proven serial integrator, not an experiment, four acquisitions digested over twelve years is precisely the capability this merger now demands in reverse^Verified · sequenceEvidence. Assembly sequence documented in the registries and brand histories: Wavefront 2008; Chelsea 16 Oct 2018; EIVA 21 May 2019; 2G Robotics 4–5 Feb 2020 (rebranded Voyis Jun 2021, the rebrand is frequently mis-dated); holding company formed 1 Apr 2020, renamed Covelya 5 Oct 2021; Forcys launched at UDT Rotterdam, 7 Jun 2022. Caveat: all four acquisition prices were undisclosed; the “serial integrator” de-risking read is my inference from the record.Sources: Sonardyne, history · Companies House, Covelya Group filings.

And here is the estate itself, the catalogue the C$615M actually buys, brand by brand^Verified · specs company-statedEvidence. The catalogue below is compiled from the brands’ own product datasheets and pages (my evidence base logs each family individually); descriptions and specs are company-stated. Independent proof the estate ships inside other people’s platforms: L3Harris’s IVER4 900 carries Voyis Recon LS + Wavefront Solstice 3000 for a NATO navy; Njord Survey selected SPRINT-Nav U for its ecoSUB AUVs; Sonardyne won the UK’s first offshore-CCS monitoring contract.Company-primary cataloguesSonardyne, products · Wavefront/Forcys, Sentinel IDS · Sonardyne, SPRINT-Nav hybrid navigationSonardyne, Origin ADCP family · Chelsea, datasheets · brand pages: eiva.com · voyis.com · forcys.comIndependent platform proofL3Harris, IVER4 (Voyis + Wavefront content), Mar 2026Sonardyne, Njord ecoSUB SPRINT-Nav win · Sonardyne, UK-first offshore-CCS monitoring awardTwo internal overlaps are real and carried below in the honest imperfections: Solstice against Kraken’s own SAS; Voyis against 3D at Depth.. Almost every row is a franchise Kraken did not own on 2 March; filter it and notice how few rows compete with anything Kraken already makes.

What the price of this deal buys is not a product list; it is half a century of installed base. Sonardyne’s founder sketched long-baseline acoustic navigation in a 1962 engineering exam and built his first product at home; by 1973 the young company was positioning BP’s Forties field, and in 1975 it began the survey of the Mary Rose, the Tudor warship raised in 1982. Six generations of Compatt transponders, the first USBL-equipped vessel (1993), the SPRINT inertial line (2010) and four Queen’s Awards later, the 2024 pair of King’s Awards landed on a company that has been the reference standard in subsea positioning for most of its existence^Verified · my researchEvidence. From my six-agent brand-history pass (Companies House, brand archives, trade press; Jun 2026) plus a deeper archival/registry pass: founder John Partridge conceived long-baseline navigation in a 1962 Bristol engineering exam and built the Rangemeter from home; incorporation 1971; BP Forties positioning from ~1973; the Mary Rose survey from 1975 (hull raised 1982); six Compatt generations; first USBL-equipped vessel 1993; SPRINT INS 2010; four Queen’s Awards plus two 2024 King’s Awards. Lore-grade items (the exam, the bedroom) are company-history, held soft. One popular claim failed corroboration and is excluded: that BlueComm featured on Blue Planet II.Sources: Sonardyne, our history (company account) · Companies House records 01030381 / 01299452 (registry, in my register).

The bolt-ons carry the same pattern, decades-deep specialists, not startups. Wavefront: four defence-sonar engineers in Dorset, 2004; Sentinel through US Navy trials at NUWC Newport within three years; Solstice the mine-countermeasures side-scan of choice on General Dynamics’ Bluefin vehicles under Australia’s SEA 1778; more than 150 Sentinel systems installed, on the company’s own count^Verified · claim in partEvidence. Timeline verified via my brand-history pass: incorporated Feb 2004 in Sherborne, Dorset, by defence-sonar engineers; Sentinel launched with Sonardyne 2007 and passed US Navy trials at NUWC Newport that October, leading to the 2008 acquisition; Solstice (2011) became the MCM side-scan on General Dynamics’ Bluefin vehicles, fitted under Australia’s SEA 1778. The “more than 150 Sentinel systems installed” figure is Wavefront’s own claim, carried at claim tier, not independently counted.Sources: Sonardyne/Wavefront, Sentinel product page (vendor) · Companies House 05032241 + trade-press archive (my register). Voyis: founded as 2G Robotics in Waterloo in 2007, founder-funded, its laser scanners on the 2022 expedition that found Shackleton’s Endurance and on the full-resolution Titanic digital twin^VerifiedEvidence. Founded 2007 in Waterloo, Ontario as 2G Robotics by Jason Gillham; founder- and revenue-funded (the only outside money a token ~CA$18k research contribution); acquired by Sonardyne Feb 2020, rebranded Voyis Jun 2021. Its imaging worked the 2022 Endurance22 expedition that found Shackleton’s Endurance, confirmed in Sonardyne’s own 9-Mar-2022 release (in my register; no public URL survives, so named), and the full-resolution Titanic digital twin.Sources: Sonardyne Endurance22 release, 9 Mar 2022 (named; no link composed) · my six-agent brand-history pass (Companies House + archives).

I don’t treat heritage as a valuation input. I treat it as evidence of the thing an acquirer cannot build quickly: fifty years of navies and survey fleets that already trust the kit. It sits alongside, not instead of, the honest-imperfections ledger above, the same group carries two brands that lean on it to stay afloat.

Twelve-plus sites across two continents before the ink dried^Verified · registry-sourcedEvidence. Every pin is a registered premises from my legal-entities and facilities register, compiled from UK Companies House, the German Handelsregister, Danish CVR, Canadian and US corporate registries, and the brands’ own published addresses. Interactive; tiles load from the internet.Sources: UK Companies House, Covelya Group (12493148) · registry records logged per-site in my evidence base, explore it: every pin is a registered premises, from Mount Pearl and Dartmouth to Yateley, Rostock, Skanderborg and Waterloo.

Two nuances keep the six-brand framing honest, both carried at full weight into the bear case. The platform is materially uneven: it is concentrated in Sonardyne plus EIVA, while Chelsea is loss-making (recapitalised by ~£3.76M in May 2026) and Forcys runs on intercompany debt, two of the six brands are financially weak and group-dependent^Verified · bear-relevantEvidence. Registry iXBRL accounts show the asset-quality split: Sonardyne International + EIVA carry the group; Chelsea Technologies is loss-making and was recapitalised ~£3.76M (May 2026); Forcys Ltd runs net liabilities funded by intercompany debt. A controversy scan found no material negatives, though for a private group, absence of evidence is not evidence of absence.Source: UK Companies House, group & subsidiary statutory filings (entity-level accounts catalogued in the evidence base). And much of the defence revenue flows through prime integrators, L3Harris, Anduril, General Dynamics, MSubs, so Covelya is often the payload sub-supplier rather than the contract holder, and several headline “wins” are positioning rather than disclosed contracts^My caveatEvidence. Channel-concentration read (my synthesis): AUKUS Pillar 2, Project CABOT and some Anduril attributions are positioning or third-party attribution, not disclosed Covelya contract values. The US vehicle “Forcys Inc.” (Delaware, Sep 2023; certificates read directly) is a thin shell whose principal office and officers sit at Sonardyne’s UK HQ, a textbook FOCI profile that confirms the mitigation trigger, not the mitigation itself.Sources: Delaware Certificate of Incorporation (file 2372141) & Virginia Certificate of Authority (Entity 11596997), user-supplied primary documents in the evidence base · my analysis.

Three things flip on the closing date. Concentration dilutes: the ~45% single-customer exposure that defines standalone Kraken is modelled to fall to ~13% of combined revenue^My modelEvidence. Derived estimate, not company-stated: folding Sonardyne’s installed base and order book into the combined revenue line cuts the top customer from ~45% (audited FY2025 concentration) to roughly 13%. The input concentration is verified; the output is my arithmetic on the combined base.Source: my model; concentration input: Kraken FY2025 results / MD&A (IFRS major-customer note). The confirmed Anduril content deepens: Sonardyne is a named Ghost Shark supplier on Anduril’s own 2023 partner announcement, so Kraken is buying the publicly-confirmed piece of the programme its own name is only inferred on^VerifiedEvidence. Anduril’s Nov-2023 ten-partner Ghost Shark announcement names Sonardyne for sub-surface navigation; no Kraken part is named on the programme (batteries are corroborated separately, buyer-side). Post-close, the group holds Ghost Shark content on both a confirmed and a strongly-indicated basis.Sources: Anduril Ghost Shark supplier announcement, Nov 2023 (named-partner list, catalogued in the evidence base) · Kraken, Dive-LD sea-trials (sister-vehicle primary), Apr 2021. And the people are locked in: the SPA embeds a seller lock-up, a UK-and-Denmark non-compete, and a retention architecture paying acquired staff through 2028, the standard failure modes of a people-business acquisition, contractually pre-answered^Verified · contract termsEvidence. Lock-up and listing-maintenance covenant, a non-compete covering the UK and Denmark (Sonardyne’s and EIVA’s geographies), and retention paying key acquired staff through 2028, all read from the executed SPA. The lock-up is also the future insider-overhang the valuation section tracks.Source: Covelya Share Purchase Agreement, SEDAR+ filing 06406273 (via sedarplus.ca).

Kraken spent a decade buying the eyes and the heart. On 2 July it buys the inner ear, and the customer list that comes with it.

One more thing the record shows: the deal was telegraphed^Their words · call transcriptEvidence. Q2-2025 call (Aug 2025): capital-allocation answer named M&A as a priority targeting US + European capabilities and customers, with European urgency explicitly higher; the Covelya SPA followed on 3 Mar 2026. Management statements on a recorded call.Source: Q2-2025 earnings call (transcript). Seven months before signing, management told analysts M&A was a priority, specifically “US + Europe capabilities/customers”, and that Europe carried “a higher sense of urgency than two years ago”. Covelya is that sentence executed, not a surprise pivot.

Kraken sells into three tiers: defence primes and UUV builders embedding its sonar and batteries (HII on the REMUS 620, Anduril on the Dive-LD, Teledyne on the Gavia and SeaRaptor); navies and governments buying directly (Canada, Denmark, Poland, the US, Australia, NATO bodies); and a commercial services book for offshore energy, Ocean Infinity, Fugro, TechnipFMC and the survey majors^Verified · ordersEvidence. Tier-one fits are primary-sourced order releases (HII 2023, Teledyne 2025, Dive-LD 2021); navy customers are contract announcements; the earnings-call names anchor the commercial roster, the broader “logo wall” is company-presented and not independently confirmed.Sources: Kraken, HII MINSAS order, Nov 2023 · Kraken, Teledyne orders, Dec 2025 · Kraken, Dive-LD, Apr 2021. A recurring feature is the frenemy structure, Teledyne and HII buy Kraken content while competing at the platform layer. That validates the merchant model; it also caps how deep any single relationship can safely go.^Verified · in partEvidence. Both frenemies are on the record buying Kraken content while competing at platform level, and the commercial book is receipted: Kraken batteries were fleet-standard on Ocean Infinity’s AUVs as early as 2019. “Frenemy” is my framing, not a company term.Defence primes buying while competingHII, REMUS 620 torpedo-tube recovery & swimoutUST, Teledyne delivers four Gavia AUVs to Sweden, Feb 2026The commercial services book: receiptsGlobeNewswire, Kraken subsea-battery milestone with Ocean Infinity, 18 Nov 2019Offshore Engineer, Kraken gives Ocean Infinity AUVs an energy boostThe cap, honestlyThe same partners compete at the vehicle layer, HII has since multi-sourced a rival SAS on REMUS variants, so relationship depth is structurally capped.

Two refinements keep this honest in both directions. First, the 45% was already being diluted organically before Covelya arrived^Company claims · dated arcEvidence. The broadening was underway before the cure arrived: on the Q4-FY2025 call (16 Apr 2026) management said it had landed “another large U.S. defense company” as a battery customer, a second anchor-class name, unnamed, alongside the “very significant” XL-UUV customer; and the March 2026 $24M announcement spanned 10+ customers in five countries including three NEW defence customers. The 45% is a snapshot of FY2025, taken just before these names start printing revenue.Sources: Kraken Q4/FY2025 call (transcript) · Kraken, $24M / three new defence customers, 17 Mar 2026, the second large US defence customer management says it landed, plus three new defence names in a single March order release, mean the standalone book was broadening on its own; the deal accelerates a cure that had already begun. Second, and this is the part a concentration table hides, de-risked is not de-levered^My model · corrected dilutionEvidence. My corrected per-share dilution model: the deal is mostly cash (C$480M cash + only C$135M stock, funded by the C$402.5M receipt raise at C$8.50), so the share count rises ~25% (307.2M basic → ~385M fully diluted), not the ~100% a miscounted comparison implies. A fixed $1M SeaPower sale is worth ~$3.26 per 1,000 shares standalone and ~$2.60 post-Covelya: −20% (−17% on a basic-share basis), roughly 2.5× less dilution than the −50% the doubled-share-count version shows. Illustrative, share-count-driven; share counts from the 5-Mar-2026 prospectus supplement + my dilution reconciliation.Source: my corrected dilution model (worksheet + prospectus supplement in my evidence base). Not investment advice.: because the deal was paid mostly in cash, the share count grew ~25%, not 100%, so a $1M SeaPower win still lands at roughly 80% of its old per-share weight inside the bigger company. Anduril and SeaPower stop being the thing that could sink the story while remaining the thing most likely to send it, concentration was cured as a downside; the torque of every battery win survives.

The most mis-read figure in the Kraken file: Kraken does not report a contracted backlog. It discloses announced order intake, a selective flow of product wins it elects to publicise, with many award values never stated and the ~C$50M/yr recurring services book never announced as orders at all.^Verified · ordersEvidence. The announced flow is real but curated, three receipts spanning the size range, all primary Kraken releases. The 3-Sep-2025 release (US, Norway, Turkey customers; 10 SAS to one buyer; four UUV size classes) pre-dates and sits outside the 2026 tally.What the announced flow looks likeKraken PR, Atlas Elektronik awards Kraken $425,000 contractKraken PR, $4.8M subsea-battery contract, defence-industry customerKraken PR, $13M in SAS and battery sales, 3 Sep 2025The honest cutSelectivity cuts both ways: unannounced awards make my intake tally a floor, and also make “order intake” unauditable from outside. The 2026 flow is real and accelerating: C$35M (13 Jan) + C$24M (17 Mar) + C$28M (16 Apr) = the C$87M management cited at the FY2025 results, rising to ~C$97M by the Q1 release, plus Covelya’s ~C$135M of Q1 intake, ~C$165M YTD^Verified · reconcilesEvidence. Each component is a separately announced, linked release, and the running totals reconcile exactly; Covelya’s C$135M is verbatim in the FY2025 release and the ~C$165M update is in the Q1-2026 release, both primary.Component releasesKraken, $35M SeaPower sales, 13 Jan 2026Kraken, $24M defence orders, 17 Mar 2026UST, $28M SeaPower + SAS orders, 16 Apr 2026Kraken, Q1-2026 results (~C$97M Kraken / ~C$165M Covelya YTD), 28 May 2026The number to distrustSecondary write-ups sum the two flows into a “~C$262M combined backlog”, wrong twice over: it sums two flows (not a stock), and no company-disclosed backlog balance exists. Treat any “backlog” headline for this name with suspicion.. Conversion speed is the other half of honesty: a signed order converts over roughly three-to-six years^My registerEvidence. The 3–6-year conversion window is my register’s working figure from programme histories, Danish MCM signed 2020 with deliveries running years out; RCN RMDS (Dec 2022) structured as a multi-year “up to” ceiling; KATFISH (2016) delivering across the following half-decade. Kraken publishes no conversion metric of its own.Sources: Kraken PR, C$36M Danish Navy contract, 2020 · Kraken, FY2025 results, 16 Apr 2026, so a strong intake year does not become a matching revenue year, and defence-sonar peer Exail booked €844M of FY2025 intake, an order of magnitude above Kraken.^VerifiedEvidence. Exail Technologies reported €844M of FY2025 order intake in its results presentation (Mar 2026), peer-scale context that keeps Kraken’s intake numbers honest: the flow is growing fast from a small base.Sources: Exail Technologies, FY2025 results presentation (PDF), Mar 2026 The flow is a floor, the label “backlog” is wrong, and the direction is unambiguously up.

And the CFO’s arithmetic on the current book, in his own words^Their words · call transcriptEvidence. Q4-FY2025 call (16 Apr 2026): C$87M product orders YTD + ~C$50M services baseline mapped against the C$165–175M guide; combined (incl. Covelya) YTD orders >C$200M; management itself flagged the run-rate caveat. Middle-East demand described as “urgent operational” on the same call.Source: Q4/FY2025 earnings call (transcript): C$87M of product orders YTD plus a ~C$50M services baseline, the visible portion of the C$165–175M guide, with combined YTD orders above C$200M, offered with the honest caveat that it is “not a guaranteed quarterly run-rate”.

The 2026 tape itself^Verified · company releasesEvidence. The 2026 cadence in primary releases: ~$24M across 10+ customers in five countries including three NEW defence customers (17 Mar); ~$28M in SeaPower battery + SAS orders (16 Apr) announced the same day as results; $35M in battery sales opened the year (Jan). Order announcements are gross bookings, not revenue.Sources: Kraken, $24M defence orders, 17 Mar 2026 · Kraken, $28M battery + SAS orders, 16 Apr 2026 · Kraken, $35M battery sales, Jan 2026: $35M in January, ~$24M in March, ten-plus customers, five countries, three new defence names, ~$28M in April. Breadth is the point: this is no longer one anchor customer’s order book.

The cadence anchors on a few programmes: the 2016 Elbit KATFISH order that started the towed line^Verified · namedEvidence. April-2016 Elbit Systems order for KATFISH towed synthetic-aperture sonar, the win that started the towed line. Primary: Kraken press release, Apr 2016, held in my evidence base; I name it rather than risk a dead or reconstructed link.Sources: Kraken PR, Elbit Systems KATFISH order, Apr 2016 (named, no stable public URL); the Royal Danish Navy C$36M MCM package (2020, won over Northrop Grumman, Thales and Klein)^Verified · in partEvidence. The C$36M Royal Danish Navy MCM contract (2020) is primary-confirmed. The competitive field, beating Northrop Grumman, Thales and Klein, is management-stated, carried at claim tier.Sources: Kraken PR, C$36M Danish Navy contract, 2020 · Kraken earnings-call transcript (competitive detail, management-stated); the Royal Canadian Navy RMDS award of up to ~C$57.9M (Dec 2022)^Verified · namedEvidence. Royal Canadian Navy Remote Minehunting & Disposal System award, announced 22 Dec 2022, valued “up to” ~C$57.9M including options, a ceiling, not booked revenue. Primary: Kraken PR and the Government of Canada award record, both in my evidence base; named rather than linked.Sources: Kraken PR, RCN RMDS award, 22 Dec 2022 · Government of Canada contract-award record (named, no stable public URLs); and an 8-year framework with an international defence contractor (2020, value never stated)^VerifiedEvidence. All primary releases. One per-unit caveat: dividing the Danish C$36M by its four systems (≈C$9M/system) overstates the hardware price, the EU tender record shows the value includes sustainment running up to 17 years.PrimariesKraken, $36M Danish Navy contract, 8 Sep 2020Kraken, RCN RMDS award (up to ~C$57.9M), 7 Dec 2022Kraken, 8-year framework agreement, 23 Jan 2020Kraken, Elbit sonar order, Feb 2016. Structurally the book splits into chunky multi-year defence programs (C$10–30M blocks, prone to timing slippage, the cause of the FY2025 guidance miss)^Verified · my readEvidence. FY2025 came in at C$102.2M against a C$120–135M guide; management attributed the shortfall to programme timing, my decomposition: roughly three KATFISH systems slipping plus a ~C$8M de-scope. The two-block structure (lumpy C$10–30M defence programs over a recurring services base) is my read of the disclosed book.Sources: Kraken, FY2025 results, 16 Apr 2026 · Kraken Q4-2025 earnings-call transcript and a shorter-cycle services base that recurs without ever making headlines.

The supply map below is the discipline that keeps the story honest: the relevant figure is content-per-platform, never the whole-vehicle price. The deepest verified stack is Anduril’s Dive-LD, Kraken batteries, MINSAS-120 sonar and Sonardyne navigation on one hull (≈C$2M of content per vehicle, my estimate).^My modelEvidence. The stack itself is primary-confirmed, Kraken’s own 2021 release has Dive-LD running MINSAS-120 and Kraken pressure-tolerant batteries over 30 missions, and Anduril then bought Dive. The ≈C$2M per-hull figure is mine: battery modules plus sonar at list-price basis; Kraken never discloses per-hull content value.Sources: Kraken PR, Dive Technologies sea trials with Kraken sonar + batteries, 2021 · Defense One, Anduril buys Dive Technologies, Feb 2022 Ghost Shark carries verified Kraken batteries, with sonar strongly indicated but never confirmed by part; the ≈US$7–10M-per-hull content figure is one broker’s model^Estimates · labelledEvidence. Fits are verified; the dollar figures are models and say so: Dive-LD ~C$2M/unit is my preliminary estimate (the fit itself was verified at 2022 sea trials); Ghost Shark ~US$7–10M/unit (~C$11M across the three prototypes) is Desjardins’ reconciliation, a single sell-side model, not company disclosure. Programme headlines (A$1.7B) bundle development, manufacturing build-out and maintenance; dividing them by hull count is a modelling error.Bear-side map, equal weightSeveral marquee “Kraken platforms” carry Kraken sonar only (GreyShark, IAI BlueWhale, Cellula); others are historic wins, not standing ones (Atlas SeaCat; Ocean Infinity’s HUGIN 6000s, the current HUGIN Superior is Kongsberg-native on both battery and sonar); and some “Kraken exposure” names carry Covelya content but no Kraken at all (L3Harris Iver, GD Bluefin, XV Excalibur).SourcesKraken, Dive-LD sea-trials (fit primary), Apr 2021Desjardins via Cantech Letter (per-unit content model).

My supply-map graphic · platform fits order-verified where cited; per-hull $ content modelled where undisclosed · full 4K original opens on click.

Management’s stated pipeline grew from ~C$900M (Feb 2024) to over C$2B, behind three demand vectors: a wave of >12 mine-countermeasures programs of record coming to RFP over 2–3 years, acute Strait-of-Hormuz MCM demand against which Kraken pre-built inventory, and the XL-UUV battery funnel (one anchor customer landed, 8–10 developers in discussion)^Company claimEvidence. All management-stated (earnings calls and results releases), not independently verifiable: the pipeline total, the >12 programs, the pre-built inventory positioning and the XLUUV funnel. A pipeline is not a backlog, and the back-end-weighted conversion that produced the FY2025 guidance miss governs how much of it books.Sources: Kraken Q4/FY2025 earnings call (transcript) · Kraken, Q1-2026 results, 28 May 2026. The caveat writes itself; so does the opportunity.

Management’s own framing of the pipeline is worth quoting rather than paraphrasing^Their words · call transcriptEvidence. Q4-FY2024 call (28 Apr 2025): pipeline growth $900M→$2B explained as new programs + visibility + some pull-forward. Q2-2025 call (Aug 2025): pipeline ~$2B, ~half sensor/platform, subsea power next-biggest; 8–10 XLUUV developers, confident of signing ≥1 within the year. Management statements on recorded calls, not independently verified.Sources: Q4/FY2024 earnings call (transcript) · Q2-2025 earnings call (transcript): the jump from ~$900M to ~$2B was “new programs + clearer visibility + some pull-forward”, roughly half of it sensor/platform work with subsea power the next-biggest block, and “8–10 companies developing XLUUVs” with confidence in signing at least one more per year.

One more feature of demand deserves its own ledger line, because in 2025 the market priced it exactly backwards. The combined group’s production footprint, Canada, the UK, Germany, Denmark, plus US entities, sits inside a dense web of favourable trade and defence-procurement frameworks. On the commercial side, CETA, CUKTCA, CPTPP (which the UK joined in December 2024), the Canada–Korea and Canada–Israel FTAs and the EU’s own agreements make most operating-company-to-customer flows duty-free^VerifiedEvidence. Production in Canada, the UK, Germany and Denmark (plus US entities) reaches most customers duty-free via CETA, CUKTCA, CPTPP (UK acceded Dec 2024), the Canada–Korea and Canada–Israel FTAs and the EU’s agreements; US defence sales enter Buy-American-waived and duty-free via DFARS qualifying-country status and the 1956 Defence Production Sharing Agreement.Defence frameworks (primary)DFARS 252.225-7002, qualifying-country sources (acquisition.gov)CBSA Memorandum D8-9-3, duty-free defence-production entry (DPSA)Context & corroborationTimeline, 2025–26 US–Canada trade war (with CRS / govt items in the register)The Bloomberg PNG CN snapshot and company social-channel items supporting this row are user-supplied captures, named in the register, not linked.. On the defence side, which matters far more, Canada and the UK are both DFARS “qualifying countries”: their defence content enters the US market with Buy American waived, no price differential, and duty-free entry under the 1956 Defence Production Sharing Agreement. That is the mechanism behind management’s line that Kraken serves US customers “without geographical challenges”^Company claimEvidence. The phrase “without geographical challenges” is management’s own characterisation of its US market access, logged from company channels. The underlying duty-free mechanism (DFARS qualifying country + DPSA) is verified, see the previous card; the framing of it is the company’s.Source: Kraken management commentary via company channels, user-supplied captures logged in the evidence base; no public permalink., and it is the structural answer to the tariff-war panic that helped compress the stock in 2025. The fear got priced; the exemption did not.

The honest read, so the edge is not oversold: qualifying-country status is table stakes among Western allies, France, Germany and others hold it too, so it screens out non-allied suppliers, above all China, rather than beating Kongsberg or Exail to a US award. The residual edges that are real are being ITAR-free (an explicit EU procurement preference under EDF and ReArm), the ability to route production across four allied jurisdictions as politics shift, and the installed base already inside allied programmes^My viewEvidence. Qualifying-country status is shared by France, Germany and other Western allies, it screens out non-allied suppliers (above all China), it does not beat Kongsberg or Exail to a US award. The residual edges that are real: ITAR-free status (an explicit EU procurement preference under EDF / ReArm), production routing across four allied jurisdictions, and installed base inside allied programmes. All are politics-dependent.Source: my synthesis over the trade-and-market-access research in the evidence base; primary framework documents linked in the preceding card.. All of it is politics-dependent, which cuts both ways, and in the current rearmament cycle is cutting Kraken’s way.^My readEvidence. That last clause is my judgement, so here is the direction-of-travel evidence for the home jurisdiction rather than a bare assertion.Canada’s rearmament: receiptsDND, Canada’s first Defence Industrial Strategy explained, Feb 2026ACOA, $16M federal defence investment in Newfoundland & Labrador, 19 Mar 2026CP24, ~$1B for domestic defence innovation and drone technology, Mar 2026What these do not sayNone of these names Kraken, the ACOA round funds five other NL firms. I cite them as procurement-environment direction, not as Kraken wins; politics can reverse.

Zoom all the way out and the demand this section describes sits inside a measurable hardware cycle. Global nuclear-submarine commissionings peaked at roughly seventy per five-year block in the late 1970s, collapsed to about ten in 1995–99 after the Soviet wind-down, and have been climbing since, roughly 28 in 2020–25, with China out-launching the United States ten to seven across 2021–25.^My reconstructionEvidence. The hull-count series is my reconstruction from open-source commissioning lists in five-year buckets, approximate by design. The China-side momentum has independent corroboration beyond hull counts.China’s underwater push: corroborationNaval News, China’s new underwater drones could threaten US West Coast, Dec 2025Army Recognition, AJX002 XL underwater drone and Taiwan Strait pressure, 2025Allied side: same doctrineBaird Maritime, XV Excalibur, Royal Navy’s large unmanned submarine testbedThe caveatCommissioning dates conflict across open sources; the claim is the shape of the curve, not any single bar. Carriers trace the same U-curve. The point is not nostalgia for big steel: the platforms are coming back with drones attached. Mainstream doctrine on both sides of the Atlantic is augmentation, not substitution, and the money now says so in appropriations language, not think-tank language^VerifiedEvidence. The augmentation money is now legislated and organised, not aspirational: the 2025 reconciliation bill wrote US$4.9B for unmanned vessels inside a 16-ship battle-force plus-up; the Pentagon’s FY2027 proposal carves ~US$54.6B for autonomous warfare (its largest-ever drone investment); Australia committed A$1.7B to Ghost Shark and Anduril’s Sydney factory delivered its first vehicle ahead of schedule; and the US Navy stood up a dedicated unmanned-acquisition office that could oversee up to 66 programs.Programs & budgetsUSNI News, reconciliation bill: US$4.9B in unmanned vessels (Apr 2025)Defense One, the Pentagon’s ~US$54.6B autonomous-warfare bet (May 2026)DefenseScoop, DoD plans largest-ever drone investment (Apr 2026)Australian Defence Minister, A$1.7B next-generation AUV programme (Sep 2025)CorroborationAnduril, Ghost Shark factory opens in Sydney, first vehicle earlyUSNI News, new Navy unmanned-acquisition office, up to 66 programs (Nov 2025)USNI News, Anduril pitches Ghost Shark XLUUV to the US Navy (Sep 2025).

The budget logic underneath the doctrine is brutal cost asymmetry. One Ford-class carrier costs about as much as three hundred Ghost Sharks; one Virginia-class submarine, about a hundred.^My calcEvidence. The ratios are my division on the per-unit chart above (Ford-class ~US$13B and Virginia-class ~US$4.3B from the CBO/CRS series; Ghost Shark ~US$40M contract-implied, an estimate). Order-of-magnitude claims only, the point survives large errors in any input.Sources: USNI News, Anduril pitches Ghost Shark to US Navy, Sep 2025 · CBO/CRS shipbuilding cost series (named; carried on the chart’s basis note, 16 Jun 2026) Navies that cannot double their hull counts can buy underwater mass by the dozen^Estimate · directionalEvidence. Third-party research houses size the same wave, one widely-syndicated forecast puts the UUV market at US$46.62B by 2035. I treat all of these as directional corroboration only.Market-size forecasts (research houses)Astute Analytica via GlobeNewswire, UUV market to US$46.62B by 2035, 7 Jan 2026MarketsandMarkets, UUV market reportVerified Market Research, synthetic-aperture-sonar marketThe caveatVendor forecasts with opaque methodologies and wide spreads; my own market model, built bottom-up elsewhere on this page, is deliberately far smaller., and every one of those unmanned hulls needs a battery section, a sonar, and seabed intelligence, which is precisely the layer Kraken sells today and Covelya deepens. The honest cross-current is on the chart’s own ledger: unmanned programmes fail like manned ones, GAO logged Boeing’s Orca US$242M over budget and three years late. Augmentation is a doctrine with a budget, not a guarantee with a schedule.

The cycle has ignition points, and 2026 lit the biggest one. Since late February the Strait of Hormuz, a fifth of the world’s oil and LNG, has been an active naval crisis, and by April it had produced a live, large-scale mine-countermeasures operation^VerifiedEvidence. The 2026 Hormuz crisis, primary-documented: Strait closed from late February; US counter-blockade from mid-April; ~95%/~99% collapse in Gulf crude/LNG shipping at the trough on a ~20%-of-world-oil-and-LNG lane; CENTCOM mine clearance under way from 11 April with two destroyers transiting; Iran reportedly unable to locate its own minefields.PrimaryUS CENTCOM release, 11 Apr 2026Commons Library briefing CBP-10636CorroborationMilitarnyi (citing NYT), Iran cannot find its minesJerusalem Post, independent carry of the NYT reportingStrongest contraryAl Jazeera visual guide (25 Jun 2026), the reported framework makes Iran clear its own minesIf the framework holds, the operational surge fades; the programmes-of-record wave below does not.. The demand signature is exactly Kraken-shaped: the binding constraint is finding mines, and the party that laid them reportedly cannot. The UK’s contribution alone put £115M of new money against mine-hunting drones and sailed a minehunting mothership loaded with autonomous sonar^VerifiedEvidence. The UK force package, from the government’s own release: drones, jets and a warship to the multinational Hormuz mission, with £115M of new funding for mine-hunting drones; RFA Lyme Bay sailed from Gibraltar on 25 May 2026 as a minehunting mothership, 100+ mine-warfare personnel, RNMB Ariadne USV, a crewless boat deploying Seacat AUVs, and REMUS UUVs.PrimaryUK MoD, UK contribution to the multinational Hormuz mission (12 May 2026)Royal Navy, RFA Lyme Bay departs with minehunting suite (25 May 2026)CorroborationBreaking Defense, how the Europe-led coalition aims to open the Strait (15 May 2026)Caveat: two KrakensThe “autonomous Kraken drone boats” in the UK’s Beehive package are Kraken Technology Group (a UK USV maker), NOT Kraken Robotics. A name collision I flag wherever it appears. Neither release names Kraken Robotics content on this deployment; the demand case here is the mission itself.. Management says the phones are ringing, and I log that as their claim, not my fact^Company claimTheir words. Kraken CEO Greg Reid, interviewed as the crisis ran (22–23 Apr 2026): the company is seeing a “rapid increase in global demand”, and the Strait-of-Hormuz challenge is “underwater mines”. Management’s framing of its own demand, logged at claim tier, not independently quantified.Sources: BNN Bloomberg, CEO interview, 23 Apr 2026 · CTV News (Bell Media), same reporting.

The second theatre is quieter and more structural: the seabed itself became critical infrastructure while nobody was pricing it. Better than 97% of intercontinental data crosses submarine cables whose disruption economics are measured in millions per hour and billions per repair^VerifiedEvidence. The seabed is critical infrastructure by the numbers: submarine cables carry >97% of intercontinental data traffic; disruption impact runs ~US$1.5M/hour (ICPC), against a US economy exposure estimated at ~US$649B (~3% of GDP, 2019 basis); the world logs 150–200 cable faults a year at US$1–3M a repair; the Nord Stream repair alone was estimated at US$1.39–1.56B.SourcesENISA (EU cybersecurity agency), Subsea cables: what is at stake (2023)CSIS, safeguarding subsea cables amid great-power competitionPolicy-research summary, cost of submarine-cable disruptions (2025)CaveatThe dollar figures are estimates from industry and think-tank work, on differing bases and dates, I use them for scale, not precision.. Since Nord Stream, the pattern, Russian survey ships loitering over cable routes, Baltic cables cut, NATO standing up a dedicated undersea-infrastructure centre and Baltic Sentry, has converted a hypothetical into standing allied tasking^VerifiedEvidence. The naval response is real and institutional: the Russian survey ship Yantar was met off Scotland by HMS Somerset and RAF Poseidons after loitering over cable routes (Nov 2025, with laser harassment of pilots reported); NATO stood up a dedicated Maritime Centre for the Security of Critical Undersea Infrastructure (2024) and activated Baltic Sentry within ~3 weeks of the Dec-2024 Baltic cable cuts.SourcesNaval News, Royal Navy responds to Yantar (Nov 2025)NATO SHAPE, CUI Maritime Centre launched (2024)CBC, Baltic cable incidents and the allied responseCaveatThe honest lag is on the record too: Nord Stream (Sep 2022) to a standing NATO centre took ~20 months. Alliances respond operationally in weeks, with procurement following slower, that is where the programme wave below comes from., and rivals are already building cable-guard AUV lines against it^Company claimEvidence. The market is already arming for cable security: EuroAtlas’s GREYSHARK AUV line, pitched explicitly to guard undersea cables, targets production of 150 units a year by 2026, in battery (Bravo) and hydrogen fuel-cell (Foxtrot) variants.Source: TURDEF, GREYSHARK unveiled to guard undersea cables (4 Sep 2025), the 150/yr is the vendor’s own forward target, accurately reported by an independent outlet; and GREYSHARK is a rival’s platform, I cite it as demand evidence, not as Kraken content.. Every hull in that build-out is a battery, a sonar and a seabed-intelligence contract in the shape Kraken and Covelya sell.

The Indo-Pacific is the theatre the budgets already believe in. The Pentagon’s own reported net assessment puts Xi’s readiness order for Taiwan at 2027^Verified · as reportedEvidence. The NYT’s “Overmatch” brief (8 Dec 2025) reports a classified Pentagon Office of Net Assessment review of US power against China, cataloguing China’s ability to attrit US fighters, ships and satellites, and reports Xi has ordered forces ready to seize Taiwan by 2027. I cannot read the classified primary; the tier covers the reporting, not the review.Source: New York Times, the Overmatch brief (8 Dec 2025); Taiwan’s response budget is explicitly asymmetric and unmanned, sea drones by the thousand, not hulls by the dozen^VerifiedEvidence. Taiwan’s answer is asymmetric and unmanned: President Lai announced an NT$1.25 trillion (~US$40B) eight-year special defence budget (26 Nov 2025) directed at asymmetric and unmanned capability; the detail includes 200,000 UAVs, 1,000 USVs and 2,000 loitering munitions. Washington added the largest-ever Taiwan arms package, ~US$11.1B across 8 packages (Dec 2025).PrimaryTaiwan Today (government outlet), President Lai announces NT$1.25T (27 Nov 2025)CorroborationUSNI News, the budget’s munitions and sea-drone detail (27 Jan 2026)Focus Taiwan, US$11.1B US arms sales, largest ever (18 Dec 2025)Strongest contraryBoth lines have wobbled since: the special budget was subsequently cut (The Diplomat, May 2026, in my register, no public link logged), and Washington paused a ~US$14B Taiwan package (WSJ-sourced, in my register). Indo-Pacific demand is real and lumpy., and the platform side of that story, Ghost Shark and the AUKUS build, is this site’s own section. The honest read: Indo-Pacific money is the biggest number and the least contracted, direction verified, Kraken’s slice unproven.

Closest to home, the programmes carry Kraken’s name or sit one door away. Canada is days from naming a builder for up to twelve submarines^VerifiedEvidence. Canada’s Patrol Submarine Programme: up to 12 boats, “tens of billions”, down to two finalists, Hanwha’s KSS-III against TKMS’s Type 212CD, with the government expected to name a supplier ahead of the July NATO summit; a KSS-III boat docked at CFB Esquimalt for the bid, and KPMG frames the programme as a ~$60B economic opportunity to 2044.SourcesGlobal News / Canadian Press, decision “in the coming days” (21 Jun 2026)Naval Technology, KSS-III at Esquimalt; 12 boats; KPMG $60B frame (27 May 2026)Janes, the 2025 downselect to TKMS and HanwhaCaveatMy register carries a Kraken–Hanwha technology link at soft grade, via Hanwha’s UUV/sonar work, NOT confirmed on the manned KSS-III, and Kraken’s bankable Canadian navy position remains the RMDS minehunting incumbency this section already logs. Either winner buys Canadian subsea content under the national-benefit lens; that expectation is my inference., while the navy’s minehunting modernisation already runs on Kraken’s RMDS. Poland’s next minehunters are assessed as a likely follow-on KATFISH buy^VerifiedEvidence. Poland’s Kormoran II minehunters already tow KATFISH: the 2020 selection came with the Polish Navy inspectorate’s on-record endorsement after sea trials, and Polish defence trade press now assesses Poland as a likely follow-on KATFISH-180 buyer for the next Kormorans, alongside Kraken’s ~US$24M March-2026 defence orders.SourcesNaval News, Poland picks Kraken for Kormoran II minehunting (11 Sep 2020)zbiam.pl (Polish defence press), likely follow-on KATFISH-180 buyer (2026)GospodarkaMorska, Gdynia naval-port sustainment supplies for the unmanned systems (Mar 2026)CaveatThe follow-on buy is a trade-press assessment, not an order, graded soft in my register; the 2020 award and the sustainment tape are the verified floor.. Denmark’s funded Navy Plan sits behind the MSF incumbency^VerifiedEvidence. Denmark’s Navy Plan under the Defence Agreement 2024–2033 commits to strengthened maritime capability, the funded national frame behind the Danish MSF-class MCM incumbency (KATFISH + ALARS since 2022) that this section logs among the landmark contracts.Source: Danish Ministry of Defence, Navy Plan (22 Apr 2025). And the loudest signal is stated American policy: a “Golden Fleet” with unmanned systems written into it in billions^VerifiedEvidence. Stated US policy, reported in sequence: the WSJ broke the “Golden Fleet” push for new warship classes (24 Oct 2025); the “Trump-class” battleship announcement followed (CBS, Dec 2025); by May 2026 the Navy was detailing how it would realise the fleet, including >US$5B in unmanned strategies and ships on the Navy Secretary’s own account.SourcesStars & Stripes, Navy details the Golden Fleet, incl. unmanned (12 May 2026)CBS News, the “Trump class” announcement (22 Dec 2025)WSJ (Seligman & Ward, 24 Oct 2025), paywalled; shared via the reporter’s X postCaveatAnnouncement-stage policy: ship classes and toplines shift between press conference and appropriation. The FY2027 ~US$1.5T request this rides on is a floated topline, not law, same caveat my budget stack carries.. Every one of these theatres converges on the same three purchases, find what is underwater, power what is underwater, know the seabed, which is the layer this order book sells.

One customer made the company. The next ten, and the merger, decide whether it stays made.

A thesis about mispricing has to explain who is on the other side of the trade. This section is that answer in three registers: the board that was rebuilt around the pivot, the insider ledger read honestly, and a share register that shows institutions have simply not arrived yet.

The stack is overseen by a board that was deliberately rebuilt for exactly this moment. Within roughly a year Kraken added Kristin Robertson (June 2025), 28 years at Boeing, including VP Autonomous Systems where she led the Orca XLUUV programme, then RTX, VADM (ret.) Michael Connor, former commander of the US Navy Submarine Force and now CEO of ThayerMahan; Kim Butler (December 2025), a CPA and career audit-committee chair; and Don Robertson (March 2026), a corporate-finance and M&A governance specialist^VerifiedEvidence. Four additions in roughly a year: Kristin Robertson (Jun 2025; 28 years at Boeing incl. VP Autonomous Systems, where she led the Orca XLUUV, then RTX); VADM (ret.) Michael Connor (former Commander, US Navy Submarine Force; CEO of ThayerMahan); Kim Butler (Dec 2025; CPA, ICD.D, career audit-committee chair); Don Robertson (Mar 2026; corporate-finance and M&A governance). Chair: Peter Hunter stepped down 16 Apr 2026; Shaun McEwan (director since 2016) succeeded him.Appointments (company-primary)Kraken, Kristin Robertson appointed to board, 4 Jun 2025Kraken, Kim Butler to board + Bernard Mills EVP Defence, 4 Dec 2025Kraken, Don Robertson to board + John Salama CIO, 30 Mar 2026Chair lineageKraken, Peter Hunter appointed Chairman, Apr 2024Kraken press release, Shaun McEwan succeeds Hunter as Chair (16 Apr 2026), logged in the register’s board-release group; link not composed from memory.Don Robertson and Kristin Robertson are two unrelated directors, distinct appointments.. The chair turned over too: Peter Hunter of Artemis Capital, chair from April 2024, stepped down on 16 April 2026 and was succeeded by Shaun McEwan, a director since 2016. Read the sequence for what it is: undersea-autonomy operating credibility, submarine-force customer credibility, audit depth for a company doubling its balance sheet, and M&A governance for the largest deal in its history, assembled in that order, ahead of need^My viewEvidence. My read: undersea-autonomy operating credibility, submarine-force customer credibility, audit depth, then M&A governance, hired in that order, ahead of need. About as credible a defence-pivot board as a Canadian small-cap could realistically assemble; Robertson’s Orca pedigree is doubly relevant because Boeing’s Orca is one of this thesis’s own competitive datapoints.The bear-side qualification, full weightA board reconstituted this quickly is unproven as a working body, and several appointees post-date the very deal they will now govern.Source: my assessment over the board releases linked in the previous card.. And the detail worth savouring: the director who ran Boeing’s Orca now sits inside the company this thesis argues will out-supply it.

The governance story starts before the board table. Karl Kenny, an ex-Royal Canadian Navy officer who had already built and run a marine-technology company, founded Kraken in St. John’s in 2012 and handed the chief-executive role to Greg Reid on 1 January 2023, in a succession planned years ahead: Reid had joined as CFO in 2015 and run operations as COO from 2019. Kenny stayed on as Executive Chairman until his death in February 2025, at 64^VerifiedEvidence. From my primary-source pass over the leadership releases and registry (an obituary and the succession releases resolved in my register): Karl Kenny, ex-Royal Canadian Navy, previously founder-CEO of Marport, founded Kraken in 2012 and handed the CEO role to Greg Reid on 1 Jan 2023; Reid had joined as CFO in 2015 and run operations as COO from 2019. Kenny remained Executive Chairman until his death in February 2025, at 64. The “planned, not forced” reading, the succession was executed two years before it was tested, is my inference, labelled as such.Sources: Kraken, leadership page · Kraken succession releases + obituary (my register; no links composed). The distinction matters for how I score key-person risk: this transition was executed two years before it was tested, not forced by events, which is why my register carries founder dependence as retired, while executive retention through the Covelya integration stays live on the watch list.

One governance line stays deliberately unpolished, because the data is soft and this document does not launder soft data. On a trailing-twelve-month aggregator view, insiders were net sellers by value, roughly 71,600 shares bought against about 977,200 sold, with CEO Reid the largest seller^DisputedEvidence. Trailing-12-month aggregator ledger: ~71,600 shares bought vs ~977,200 sold; CEO Reid the largest seller (~261,600 shares at ~C$2.88, about C$753k), plus a Bloomberg-sourced ~810,000-share six-month figure that does not fully reconcile with the per-transaction records. The aggregators conflict on at least two items: who sold 142,500 shares on 30 Aug 2024 (Shea per one source, Reid per another) and the exact date of Reid’s largest sale. None of it is verified against SEDI, the primary insider-disclosure system, so I quarantine the ledger from any load-bearing use until it can be verified against SEDI.Sources (named; no single public permalink logged): INK / Globe insider data; TipRanks; MarketBeat; Canadian Insider; Yahoo Finance; Bloomberg screens, user-supplied captures in the evidence base.. None of that ledger is verified against SEDI, the primary insider-disclosure system; the aggregators conflict with each other on at least two entries. Until it reconciles, it stays quarantined: logged, displayed, never load-bearing.

What can be said fairly: insiders bought the January 2023 dip, then trimmed into a re-rating that made the stock a multi-bagger, ordinary profit-taking on the pattern, while aggregate insider ownership held broadly steady around 10% on the aggregators’ basis and Reid still holds about 7.36M shares (~2.4%)^My viewEvidence. The fair characterisation: insiders bought the January 2023 dip, then trimmed into the 2025–26 re-rating, ordinary profit-taking on the pattern; aggregate insider ownership held broadly steady around 10% on the aggregators’ basis, and Reid still holds about 7.36M shares (~2.4%). A verdict either way waits on the SEDI reconciliation.Source: my read over the disputed ledger in the first card of this passage.. On the institutional side, the September 2025 holder screen shows Mawer (~9.3%), Penderfund (~7.7%) and FMR (~3.3%) at the top, institutions at roughly 26%, and short interest negligible at ~0.5%^Single sourceEvidence. Bloomberg holder screen (filings to 30 Sep 2025): Mawer ~9.28%, Penderfund ~7.70%, FMR ~3.3%; institutions ~25.8% overall; short interest negligible at ~0.46%.Basis note (labelled, per the register)A fresher 30-Apr-2026 filings-only screen computes insiders ~3.24% and institutions ~24.3%, a different basis (filed positions only) from the aggregators’ ~10% insider figure. The two bases are never comparable one-to-one and are never co-rendered here without these labels.Source: Bloomberg L.P. ownership screens, user-supplied terminal captures; no public URL.. A noisy seller’s ledger is a flag to keep watching, not a verdict, the people running the company kept their tenth of it through the entire re-rating.

The share register is the mispricing argument drawn as a table. Institutions hold about a quarter of the company, insiders ~3.24% on the same filed basis, and retail roughly 72%^VerifiedEvidence. 30-Apr-2026 filings-basis screen: institutions ~24.3% (74.7M sh), insiders ~3.24% (9.95M sh), retail ~72.4%; top-25 holders = 26.78% of the register; the US 13F line is thin at ~3.57%.Ownership screensFintel, KRKNF institutional ownership (13F/NPORT line)Simply Wall St, Kraken Robotics ownership breakdownSame basis rule as the insider ledger above: this is a filings-only basis. Aggregator insider figures (~10%) use a different basis and are never co-rendered here without that label.. There is still no institutional-scale Kraken product anywhere in the ETF complex, the one dedicated fund in the SEC pipeline is a 2× daily-leveraged retail instrument on the ADR line, an amplifier of the register’s character rather than a counterweight to it. Where conviction exists, it is specialist and concentrated: one hedge fund carries Kraken as its largest position at a verified 14.3% of NAV^Disputed · both recordedEvidence. Deep Sail Capital Partners carries Kraken as its largest single position: 14.3% of NAV verified against the Q1-2026 letter (31-Mar-2026). An undated user-supplied screenshot shows 17.2%, recorded as disputed until a dated primary resolves it.SourceBuyside Digest, Deep Sail Capital Partners fund profileThe 17.2% screenshot is user-supplied, named in the register, not linked.; Mawer’s closed New Canada fund holds ~C$43M at 3.4%; and the entire thematic-defence-ETF bid tops out at a 1.80% weight in a single WisdomTree basket, with the largest dollar position (~US$17.35M in Global X’s SHLD) amounting to 0.23% of that fund. The sharpest contrary datum gets its full weight rather than a footnote: Pender, early, capable, small-cap-native, sold roughly sixty percent of its firm-wide position into the run and dropped out of its own top-10^VerifiedEvidence. Fund-level positions, as filed or published: Mawer New Canada ~C$43M / 3.4% of the fund (5th-largest; fund ~C$1.27B, closed to new retail; 31-Mar-2026). Fidelity small-cap complex ~6.9M sh ≈ 2.24% of Kraken firm-wide (NPORT, Jul–Sep 2025). Thematic ETFs: Global X SHLD largest in dollars (~US$17.35M) at just 0.23% of fund; WisdomTree WDRN the highest weight anywhere at 1.80% (15-Jun-2026 snapshot). One dedicated fund is registered: the Defiance “Daily Target 2X Long Kraken Robotics ETF” (ticker KRK, Tidal Trust II; 485APOS filings 15-Apr & 7-May-2026, 497 filing 26-May-2026) seeks 200% of the DAILY move in the KRKNF ADR, verified against the SEC filings, correcting my earlier name-collision reading (the collision hazard with Kraken the crypto exchange, Payward Inc., persists in screeners and headlines). Daily-reset leverage decays path-dependently in both directions; it deepens the register’s retail tilt rather than offsetting it.Fund pages & filingsMawer, New Canada Fund holdings pageFintel, NPORT-derived US mutual-fund positionsThe contrary file: full weightPender, an early and capable small-cap holder, cut firm-wide from ~23.6M to ~10M shares and fell out of its fund’s top-10 (4.1% at 31-Dec-2025 → <3% by 29-May-2026). A manager who did the work sold most of the position into the re-rating.The TradingView/FactSet ETF-holdings snapshot (15-Jun-2026), the Pender MRFP/fund page and the SEC 485APOS name-collision filing are recorded in the register without stable public deep-links, named, not linked.. My read is that this table is the thesis in miniature: the re-rating to date happened without the institutional bid, and the register, thin 13F line, no index products, retail setting the marginal price, is what a mispricing looks like from the inside^My viewEvidence. A stock does not re-rate five-fold with an ownership map this empty unless most of the buyers have not arrived. Under-ownership is read here as both the mechanism of the mispricing (no index bid, thin 13F line, retail-set marginal price) and the shape of the future bid (post-close index and ETF adds, deeper institutional coverage). Falsifiable: watch the fund ledger, if institutions stay away through the integration year, this read is wrong. The cut also runs the other way: a ~72% retail register means flow-driven drawdowns, like the one being marked to market this quarter.Source: my synthesis over the Fund & ETF Holdings ledger in the evidence base (basis 15-Jun-2026)..

Companies confess in job ads what they won’t say in releases^My read · job-ad forensicsEvidence. My two-ATS snapshot (14 Jun 2026): Kraken’s Rippling careers board (29 open roles) plus each Covelya brand’s careers page, read as operational signals. Grade soft by design, job ads are intent, not results, and each inference is labelled in my register.Source: my hiring-behaviour register (ATS snapshots logged in my evidence base); inference tier.. Kraken’s board is 62% operations, a three-hub manufacturing ramp (Dartmouth, Mount Pearl, Rostock) hiring in parallel, plus a supplier-quality build-out and the single strongest tell: a Senior ILS Analyst for full military lifecycle support (MIL-STD-1388 tooling), which is what you hire when you expect to sustain a programme of record, not ship one-off demos. A US-person SAS software team is being stood up remotely, the quiet US-localisation tell, and the ads confirm RTSAS runs on NVIDIA Jetson, exactly the SWaP-C story the sonar moat rests on. On the Covelya side: Sonardyne hiring an acoustic-signatures scientist (next-gen defence R&D), Chelsea physically consolidated into Blackbushe, EIVA hiring off a record 2024 profit, and one job spec already spans “Sonardyne, Forcys & Chelsea”: the integration visible in the recruitment pipeline before the deal even closed.

The coverage map is small, young and polarised^Aggregator-dated · conflicts notedEvidence. Coverage actions compiled from financial-data aggregators (dates approximate, grade soft) plus primary underwriting releases. Conflict flag carried: Desjardins, the street-high target, was also lead underwriter on the 2025 and 2026 raises; the lone Underperform is N. Boychuk at ATB, the analyst previously associated with Cormark coverage of the same name.Sources: TheFly, ATB downgrade to Underperform C$6.50, Mar 2026 · TheFly, National Bank upgrade to C$13.00, Mar 2026 · Cantech, Desjardins C$14.00 · consensus ~C$10.80, and its arc tracks the thesis: targets chased the order flow up all through 2025, split violently on the deal, and consensus now sits ~70% above the last close.

The bull case is logged the way the bear case is: a register of sixty-four discrete drivers, each carrying a conviction grade and a paired bear counter. The most important feature is stated up front: only three clear the CONFIRMED bar, the depth-proven pressure-tolerant battery technology, the fully-funded balance sheet, and the genuinely owned sensing IP in sub-bottom sonar and LiDAR^My registerEvidence. The driver register is my structured grading of the logged evidence; its three CONFIRMED entries rest on primary sources (fielded 6,000 m batteries; FY2025 cash C$120.5M; office-recorded patent assignments). The majority of drivers are demand-tier or optionality, ecosystem, not booked Kraken revenue, and are graded OPEN until conversion.Sources: my evidence register · Kraken, FY2025 results (cash) · US12585019B2 (owned-IP example). The bull case is broad and real, but front-loaded on demand and optionality, and back-loaded on the conversion and integration proof that would crystallise value. That asymmetry is the whole investment debate.

Demand: world military spending hit records in 2024–25 and NATO’s Hague pledge set a 5%-of-GDP path, with seabed warfare and mine countermeasures pulling directly through Kraken’s lines^VerifiedEvidence. SIPRI records ~US$2.7T world military expenditure (+9.4% real, 2024); NATO’s Jun-2025 Hague declaration sets the 5% path; the counter is logged too: macro demand is not Kraken’s captured share.Sources: SIPRI, world military expenditure · NATO, Hague Summit declaration. The deal as a cure for concentration: ~45% single-customer exposure diluted to a modelled ~13%.^My modelEvidence. The ~13% is my arithmetic, not a disclosure: ~45% of FY2025 revenue (~C$46M of C$102.2M) from the largest customer, held flat over an illustrative ~C$365M combined top line. The combined base is my construction from the deal release’s figures; management has published no pro-forma concentration number.Sources: Kraken, FY2025 results, 16 Apr 2026 · Kraken, Covelya acquisition release, 3 Mar 2026 Deal arbitrage: a ~24%-margin, ~41%-EBITDA-growth asset bought at ~10–12× EBITDA into a vehicle trading at ~20× at the 30-Jun close, multiple-accretive if integration holds.^My calcEvidence. Covelya’s ~24% EBITDA margin and ~41% EBITDA growth are company-stated deal-release figures; the ~10–12× entry multiple is my arithmetic; the vehicle multiple is ~20× run-rate at the 30-Jun close of C$6.34 (~23× at the mid-June C$7.12 snapshot). Caveat: third-party EBITDA readings for the vehicle span ~17–26×, and the arbitrage only crystallises if the market awards Kraken’s multiple to the acquired earnings.Sources: Kraken, Covelya acquisition release, 3 Mar 2026 · StockAnalysis, PNG.V (price basis, dated) Re-rating mechanics: TSX graduation and eventual index eligibility lift a structural ownership cap (institutional ownership sits under ~30%)^Verified · mechanismEvidence. TSXV listing excludes the S&P/TSX index families and many mandates; graduation mechanics are procedural once filed (recent graduates cleared in ~2–2.5 weeks). The moderator’s flag survives onto this card: graduation removes a barrier, it does not create earnings, and an internal graduate study is a small hand-built sample, illustrative only, clean sample of 8: mean ~2.5× but median ~0.84× (below breakeven; the mean is outlier-driven), 62.5% win rate; extended sample of 11: mean ~1.53×.Sources: TMX, TSXV→TSX graduation · my graduation-cohort study (clean n=8 / extended n=11; reliability C, selection-bias caveat). And the de-risked close: the deal’s one true binary cleared on 18 June.^VerifiedEvidence. All regulatory and stock-exchange approvals received, announced 18 Jun 2026; close expected 2 Jul 2026, subject to customary closing conditions. As of my check on 2 Jul 2026 the closing release itself had not yet been published, approval is confirmed, completion is not.Sources: Kraken, regulatory approval of Covelya acquisition, 18 Jun 2026

The five pillars above are the distillation, this is the whole register^My register · gradedEvidence. All 64 drivers of my bull register, each carrying (in the underlying register) its evidence trail, its bear counter, and a written kill-criterion, the condition under which I strike it. The grading is the discipline: only 3 are CONFIRMED against primary sources; STRENGTHENING means the evidence has moved my way since logging; ACTIVE means live and evidenced but not primary-proven; OPEN means logged with a credible basis awaiting confirmation; LOGGED means catalogued at lower conviction. A bull case that grades itself this hard has earned its conviction where it claims it.Source: my bull register (worksheet in my evidence base; drivers individually evidenced). Statuses as graded at 1 Jul 2026.. Filter by grade and read the shape of it: the case does not rest on any single row, the highest-conviction rows are the ones already banked, and every row has a written kill-criterion.

The single most under-rendered part of my research was the budget stack, so here it is in full^Verified · primary budgetsEvidence. Every figure from budget documents, statutes or SIPRI/NATO primaries: world military spend a record US$2,718B in 2024 (+9.4% real, steepest since the Cold War), est. US$2,887B in 2025, NATO ~US$1,506B (55%); the Hague 5%-of-GDP-by-2035 pledge (3.5 core + 1.5 related, national roadmaps due mid-2026); US FY2026 ~US$1.01T (+13.4%) with the FY2027 request floated at ~US$1.5T, the highest single-year topline since WWII (heavily reconciliation-funded, that caveat matters); FY2026 NDAA signed at US$901B (~$8B above request); EU ReArm/Readiness-2030 up to €800B incl. the €150B SAFE facility; Canada at 2% (~C$63B, first time since ~1990) framing a “half-trillion-dollar decade”.ToplinesSIPRI, world military expenditure · NATO, Hague 5% declaration · CSIS, unpacking the $1.5T FY2027 topline · FY2026 NDAA $901B signedEurope & CanadaEC, Readiness 2030 (€800B) · EU Council, Canada joins €150B SAFE · Canada DIS: underwater autonomy a Key Sovereign CapabilityMy standing caveat survives here: a rising backdrop raises the probability and size of relevant procurement, it does not itself convert to Kraken revenue, and management expects the majority of revenue to remain export.: a record world topline, the NATO 5% pledge, a US budget heading from $1.01T enacted toward a floated ~$1.5T FY2027 request, and inside it, the autonomy slice growing at multiples of the whole^Verified · the autonomy sliceEvidence. The slice aimed at Kraken’s market is growing faster than the whole: the US Navy’s FY2026 request put ~US$5.3B to unmanned/autonomy (+~US$2.2B YoY) inside a ~US$13.4B DoD autonomy total; the FY2025 reconciliation bill added ~US$4.9B for unmanned vessels, ~US$1.3B UUVs, ~US$2.1B medium USVs, plus dedicated payload lines (ASW sonar, autonomous underwater munitions); and the Navy stood up a dedicated Robotic-Autonomous-Systems acquisition office consolidating up to 66 programmes across six PEOs, a structural commitment, not a cycle.Sources: USNI, reconciliation $4.9B unmanned vessels · USNI, Navy RAS office (66 programmes). Money at the top of a funnel is not revenue at the bottom; but the funnel has never been this full, and every layer of it is primary-sourced.

Every rung is double-edged, the same events that could re-rate the stock are, if they slip, the precise contents of the bear case^Verified · demand signalsEvidence. The in-service signals are primary: record SAS adoption at NATO REPMUS 2025; AUKUS’s first Pillar-II UUV project (May 2026); Canada’s 2026 Defence Industrial Strategy naming uncrewed underwater systems among ten Key Sovereign Capabilities (categories, not companies, the Kraken benefit is inference).Sources: Kraken, REPMUS 2025 · CNN, AUKUS seabed-cable project, May 2026 · Kraken, Q1-2026 results.

Here is the driver I think the market prices worst, because it is boring: the share register. As of 30 April 2026 Kraken was roughly 72.4% retail, ~24.3% institutional and ~3.2% insider, a register I mapped myself with a three-agent fund-ownership sweep (three parallel searches across SEC N-PORT and 13F filings plus holdings aggregators, cross-checked against each other)^Verified · registerEvidence. My three-agent fund-ownership sweep (parallel N-PORT/13F + aggregator searches, cross-checked) put the 30-Apr-2026 split at ~24.3% institutional / ~3.2% insider / ~72.4% retail. Caveat carried: the headline institutional-ownership figure is effectively single-provider at the aggregator level, so I treat the precise percentages as approximate; the retail-heavy shape is not in doubt.Sources: Fintel, KRKNF institutional holdings · my fund-ownership sweep (methodology above). The thematic money that should own a defence-tech pure-play barely does: the largest dollar position among ETFs is Global X’s SHLD at ~US$17.35M, a 0.23% weight, and the highest weight is WisdomTree’s WDRN drone fund at 1.80%.^Verified · my registerEvidence. Positions from my fund-and-ETF holdings register, built with a three-agent sweep of SEC EDGAR N-PORT/13F filings cross-checked against holdings aggregators (Jun 2026): Global X SHLD ~US$17.35M at a 0.23% weight (largest dollar position), WisdomTree WDRN at 1.80% (highest weight). Holdings data as of Apr–May 2026 vintages; ETF weights drift daily.Sources: Fintel, KRKNF institutional ownership · SEC EDGAR N-PORT/13F filings (my register; per-fund filings not individually linked) My read is that most of the natural buyers have not passed on Kraken; they structurally cannot buy it. A TSX Venture listing excludes the stock from the S&P/TSX index families and from a long tail of institutional and passive mandates that require a senior-board listing^Verified · the capEvidence. Exclusion is mechanical, not judgemental: S&P/TSX index methodology requires a TSX listing, and many mandates key off index membership or senior-board status. The graduation route itself is procedural once filed, board-approved application plus TSX review, no shareholder vote absent a paired consolidation, and recent graduates moved from conditional approval to senior-board trading in roughly 2–2.5 weeks.Sources: TMX, TSXV→TSX graduation · TSX, dual-listing guide (US-listing mechanics) · S&P Dow Jones Indices, S&P/TSX methodology.

The mechanism, weighed honestly. Graduation lifts the cap; index eligibility under the S&P/TSX methodology and its periodic reviews then opens the passive channel, the documented “index effect” of inclusion-driven inflows. But I will not sell that mechanism harder than the evidence sells it. The strongest contrary source is academic: Greenwood & Sammon’s “The Disappearing Index Effect” (NBER working paper) finds the inclusion pop has shrunk toward zero over recent decades.^VerifiedEvidence. Greenwood & Sammon, “The Disappearing Index Effect” (HBS/NBER working paper 23-025): the average price pop from S&P 500 index inclusion has declined toward zero in recent decades. I cite it against my own driver on purpose, it attacks the magnitude of inclusion flows, not the mechanism of mandate eligibility. Named without a URL rather than risk a wrong one.Sources: S&P DJI, S&P/TSX index methodology · Greenwood & Sammon, HBS/NBER WP 23-025 (named, no link composed) And my own graduation-cohort study, a hand-built convenience sample of TSXV→TSX graduates, reliability-graded C with a selection-bias caveat, is honest about its middle: clean sample of eight, mean total return ~2.5× but median ~0.84×, below breakeven; the mean is outlier-driven^My model · mechanism vs sizeEvidence. The exclusion/eligibility mechanism is verified; the size of the prize is my inference and the evidence cuts both ways. For: index-family eligibility plus mandate un-capping on a register that is ~72% retail. Against (carried at full weight): the index effect has largely disappeared in the academic record, and my own graduate study’s median outcome is below breakeven, graduation removes a barrier, it does not create earnings.Sources: S&P DJI methodology · Greenwood & Sammon, “The Disappearing Index Effect” (NBER working paper; named, no public link logged) · my graduation-cohort study (clean n=8 / extended n=11: mean ~1.53×; methodology and caveats as stated). So I hold this driver the way the register holds it: mechanism confirmed, magnitude open.

What I am actually watching from 2 July. First, a graduation application, management has flagged the senior board as an objective, but there is no committed, dated corporate event yet, and I refuse to let paid-IR “uplisting” chatter stand in for a filing. Second, once senior-board, the S&P/TSX review cycle, eligibility is assessed on the exchange’s published methodology, not on narrative. Third, ETF-weight evolution: whether SHLD and WDRN scale their positions as float and liquidity grow post-close, and whether new defence-tech products add the name. Fourth, the N-PORT/13F tape: whether new long-only names appear beside Mawer, and whether Deep Sail’s conviction position holds.^Verified · in partEvidence. From my N-PORT/13F sweep: Deep Sail Capital held Kraken as its largest position at 14.3% of NAV (verified 31 Mar 2026), a higher 17.2% figure circulates and is disputed in my base, so I carry the lower, verified number. Mawer New Canada holds ~3.4% (fifth-largest; fund closed to new retail); Pender trimmed Kraken out of its top ten by May 2026.Sources: Fintel, KRKNF institutional ownership · SEC EDGAR N-PORT/13F filings + fund letters (my register) One instrument needs naming precisely: the Defiance “2X Long Kraken Robotics” filing (ticker KRK, Tidal Trust II) targets 200% of the daily move in the KRKNF ADR line, real, SEC-filed, and this company. It amplifies the register’s retail character rather than counterweighing it, and the name collision with Payward’s crypto exchange remains a screening hazard, not a doubt about the referent^My read · the watch listEvidence. The watch list is mine; each tell is checkable on a public tape (TSX bulletins, index-review announcements, ETF holdings files, N-PORT/13F). The flow argument is my inference, and it is two-sided: passive money sells as mechanically as it buys, a deletion, a weight cut or a derate runs the same machine in reverse. The KRK reading comes from the registrant’s own SEC filings, Tidal Trust II / Defiance 485APOS (15-Apr & 7-May-2026) and 497 (26-May-2026): 200% of the daily KRKNF move, daily-reset leverage decaying both directions. Payward’s crypto exchange trades under the same “Kraken” name, a collision hazard for screeners, not the fund’s referent (filings named; no public URL logged, none composed).Sources: Fintel, KRKNF holdings tape · SEC EDGAR, Tidal Trust II 485APOS/497 (Apr–May 2026; named, no link) · my fund-ownership sweep. If the flip happens, it will be visible in exactly these four places, and if it has not started showing up within a few quarters of a graduation, that silence is bear-case evidence and I will log it as such.

At the 30-Jun close Kraken is a roughly C$2.4B company still quoted on Canada’s venture exchange, an anomaly of scale that management intends to fix: graduation to the senior TSX is pencilled in for the second half of 2026, gated on the Covelya close^Company intent · corroboratedEvidence. Graduation to the senior TSX is “pencilled in for the second half of 2026”, gated on the Covelya close, with CEO Greg Reid openly frustrated it has not happened already (without the deal on the table, he says, Kraken would already check every TSX listing box). That is independent financial media corroborating the same intent my register first received through Kraken’s paid IR channel (Sophic Capital emails, Oct-2025), two channels, one fact. My discipline stands: no graduation application has been filed or dated, so this stays a forward-looking company intent, not an event.Weight it honestlyNAI500 is independent of Kraken’s paid IR, but it sells IR services itself, so I grade it soft: it corroborates the stated intent, not that it will occur.Sources: NAI500, “Kraken Robotics Eyes TSX Upgrade, Is the Window Still Open?”, 23 Apr 2026 · Sophic Capital IR emails, Oct-2025 (my register; no public link). The mechanics are the easy part, and I have verified them separately: a board decision and a listing application, no shareholder vote, requirements Kraken clears comfortably on my reading of the published tests, and a public phase recent graduates have run in about two weeks^Verified · mechanicsEvidence. Graduation is a voluntary four-step process (advisory meeting → TSX Listing Application → conditional approval → trading) requiring a board decision but no shareholder vote absent a paired consolidation, which Kraken does not need. Recent graduates ran the public phase in ~2–2.5 weeks (IsoEnergy: conditional 21-Jun-2024 → TSX trading 8-Jul-2024; Magna Mining reached conditional approval May-2026). At ≥C$100M market cap the application fee is waived and escrow exempted; ticker and CUSIP carry over. “Comfortably clears” the Original Listing Requirements is my assessment against the published tests (FY2025 revenue C$102.2M against a C$10M revenue test; ~C$2.4B cap), not an exchange determination.Sources: TMX, TSXV-to-TSX graduation guide · Magna Mining, conditional graduation approval, 4 May 2026 · my registry pass over TMX graduation rules + six recent graduate PRs (IsoEnergy, Tiny, NMG, Volatus, Hemlo, Magna; dates from the releases).. What deserves more space than the mechanics is what leaving the venue is actually worth, because 2026 has been a live demonstration.

The venue Kraken trades on has spent the year falling while the board it would join kept setting records. The Venture Composite was −9.4% for the half in late June while the senior Composite tracked toward its eighth consecutive positive quarter, and the pattern is not a 2026 novelty: over five years the senior board compounded ~+74% while the Venture went backwards, and in drawdowns the junior board swings roughly half again as hard. Every one of those venture-bucket selloffs has hit Kraken’s tape regardless of what the business did that week:

So what does escape buy? My honest answer: a real but moderate de-risking, and I have apportioned it. Graduation changes who can own the stock, mandate eligibility, tighter spreads, and eventually the passive channel once index inclusion follows, and it strips out the venture-bucket selling that hits every junior name in a risk-off tape. It does not touch fundamentals, contract lumpiness or customer concentration; my read is that only ~20–40% of the volatility relief graduates enjoy is the venue itself, the rest being the maturity that qualified them^My readEvidence. The verdict of the senior-vs-venture study I commissioned (public index/price-return series plus mandate and index-methodology reading, 12-Jun-2026): a TSXV→TSX move is a real but moderate de-risking, it changes who can own the stock (mandate eligibility, liquidity and spreads, the passive channel after index inclusion) and strips “venture-bucket” selling, but leaves fundamentals, contract lumpiness and customer concentration untouched. My apportionment: roughly 20–40% of the post-graduation volatility relief is the venue itself; 60–80% is the company having matured enough to qualify. Index inclusion lags graduation by ~6+ months on the S&P/TSX review cycle, and Composite entry has its own float-adjusted bar (~C$2.0B) that Kraken sits near, not safely above.Source: my commissioned study (methodology as stated; research file, no public URL) · S&P DJI, S&P/TSX index methodology. Estimate; not investment advice.. The register’s framing survives: this driver removes ceilings, it does not create earnings.

And I refuse to price the graduation on folklore, so I built the base rate myself, a roster of recent TSXV→TSX graduates and what actually happened to them:

The roster’s lesson is uncomfortable and useful at once. The average graduate made money; the median graduate did not beat a coin flip, and the winners have one thing in common that has nothing to do with the listing venue: they graduated into a bull cycle for their sector, while the losers graduated into a topping one. That is the honest frame for Kraken: the act of graduating is worth little by itself, but Kraken would be graduating into the strongest naval-rearmament cycle in a generation, the miners-into-gold analogue, not the tech-into-2021 one, with the venue drag documented above removed on top^My inferenceEvidence. The cohort read is mine: the graduates that made money graduated into a bull cycle for their sector (2023-24 miners into gold); the ones that lost graduated into a topping one (2020-21 tech). Kraken would graduate into the strongest naval-rearmament cycle in a generation with a defence-tech premium attached, the favourable analogue. But my own register grades this driver modest and mechanical (graduation removes ceilings; it does not create earnings), and the magnitude stays OPEN until flows actually show up.Source: my read across my graduation-cohort study + my bull register grading. Inference; not investment advice.. What I am watching is unchanged and mechanical: an application press release, conditional approval, first TSX trade, then the S&P/TSX review cycle roughly six months behind it.

The hardest-proven parts of the bull case were never in dispute. The contested part, conversion, is what the next four quarters exist to answer.

And three drivers I have deliberately not spent yet^Logged drivers · gradedEvidence. Three drivers from my graded bull register, rendered with their own break-conditions: (1) capacity build-out, a new US-Northeast battery line (~$200M/yr capacity, end-2025) plus the Nova Scotia line and Rostock IP transfer; breaks if orders stay tens-of-millions while capacity sits idle. (2) Repeatable-expansion proof, Dartmouth is the first major line stood up outside Germany; n=1, learning curve unproven. (3) ‘Best-of-both-worlds’ Canada positioning amid 2025–26 US-alliance turbulence, real and geopolitically tail-winded, but two-sided by definition.Source: my bull register (drivers individually evidenced and graded in my evidence base; capacity figures company-stated).: the US-Northeast battery line (~$200M/yr of capacity by end-2025, the future US site the market hasn’t priced), the proof that Kraken can now replicate its manufacturing outside Germany, and the Canada positioning that sells to both sides of a fragmenting alliance. Each is graded, each carries its own break-condition, none is required by the base case.

This section is owned by the thesis’s honesty function and carries the same standing as the bull case: a register of 40 bear points (23 active, 17 retired but retained), each graded for severity and paired with its bull rebuttal. A weakened bear point is not deleted here; it is shown with its status. [My register]^My registerEvidence. What this is: my bear register, a living table in my evidence base, kept alongside the bull case since the project opened, one row per failure mode, each graded low/medium/high severity, paired with its bull rebuttal, and re-graded as filings and releases land. Weakened bears are retired in place, never deleted, so the count (40 entries: 23 active, 17 retired) is auditable. Entries logged after the manuscript froze appear below at full weight.Source: my bear register (evidence base), my own compilation and grading; no public document.

Two structural bear points deserve their own sentence. The certification gap: after a deliberate deep dig, no granted NAVSEA S9310/LBSP battery authorization for SeaPower could be affirmatively confirmed anywhere public, the company’s own language is “designed for compliance”, while a General Dynamics pack has held Navy safety approval since 2013^Verified · negative findingEvidence. A negative finding, strengthened on re-search: Kraken’s public claim is compliance-by-design plus NSWC-Carderock CRADA testing and fielded use; no public certificate exists. Mitigant, at equal weight: S9310 approvals are application-specific and typically initiated by the platform integrator, so absence is not proof of failure, but the gap is real, and a competitor holds the credential.Sources: Kraken, US Navy CRADA, 2021 · GD, Bluefin battery Navy safety approval, 2013. And the energy-substitution boundary: fuel cells and diesel-electric beat Li-ion for weeks-long missions, and a senior RCN officer publicly flagged that batteries “diminish in cold water”, Kraken publishes no derating curve^Verified · boundedEvidence. VAdm Topshee (Commander RCN) on cold-water battery performance and Arctic missions; the substitution physics is covered in the Technology section, real, rising, and mission-bounded to the extreme-endurance tier, with the fuel-cell buffer pack keeping Kraken content in the vehicle.Sources: Politico, VAdm Topshee interview, 13 Feb 2026 · my technical synthesis (appendices). Guidance-quality context belongs on the same card: the FY2025 miss was timing (units slipped, one component de-scoped ~C$8M to 2027), but a miss is a miss^VerifiedEvidence. FY2025 guidance C$120–135M reiterated through Q3, actual C$102.2M (−16–24% vs midpoint); organic −8% derivation strips 3D at Depth’s audited C$18.2M inorganic contribution (my arithmetic on audited segments); FCF ~−C$29M from the cash-flow statement.Sources: Kraken, FY2025 results, 16 Apr 2026 · Q4/FY2025 earnings call (transcript).

The register kept working after the thesis text froze, and honesty requires the entries it added since to sit here at full weight, not in a drawer. Cross-adversary exposure. The same merchant neutrality the bull case celebrates cuts both ways: Kraken is deepening a KATFISH relationship with Turkey’s SEFINE, operator of NATO’s largest mine-countermeasures fleet, while its first big order came from Elbit Systems and Covelya’s brands sell across a customer map that spans rival capitals^Verified facts · register framingEvidence. The relationships are verified; the forced-choice scenario is the register’s inference. KATFISH advanced from a Q1-2026 Istanbul demonstration of autonomous launch-and-recovery off SEFINE’s RD-22 USV to a 6 May 2026 MOU folding KATFISH into SEFINE SISAM’s mission-planning and target-recognition software; Kraken’s first large order (>C$15M, 2016) was Elbit’s Seagull USV. An MOU is not a contract, and this exposure is political, not contractual: if export-control politics ever force a choice, neutrality becomes a decision.The recordKraken, SEFINE SISAM MOU, 6 May 2026 (GlobeNewswire)Kraken, MOU release (company)Naval News, KATFISH auto-LARS demo from SEFINE USV, Apr 2026Register entryDual-use / cross-adversary exposure, graded low–medium, my bear register (evidence base, post-freeze); user-supplied, no public link..

Goodwill on the balance sheet. C$615M of consideration against a target with modest tangible assets re-loads the balance sheet with goodwill and stepped-up intangibles. The prospectus itself warns that amortisation of those intangibles plus new interest expense will pull reported net income well below what combined Adjusted EBITDA implies, and if integration runs slower than guided, the same accounting delivers an impairment test and a headline-ugly, non-cash writedown^Register · prospectus-anchoredEvidence. The prospectus supplement (5 Mar 2026) is explicit that its “Combined Financial Information” excludes purchase-price allocation and fair-value step-ups, and that post-close amortisation and interest will land below the Adjusted-EBITDA line. The impairment pathway is the register’s extension (graded low–medium), not a company statement.Sources: Prospectus Supplement, 5 Mar 2026 (SEDAR+, filed under Kraken Robotics Inc.) · my bear register (evidence base).

The budget cycle turning. The demand thesis leans on a rearmament supercycle; a US continuing-resolution freeze or a durable peace re-pricing would compress order flow and the multiple simultaneously, and the tape has already demonstrated the mechanism, marking the stock down ~34% from its March 2026 high on a ceasefire without a single company misstep^Register · medium over a cycleEvidence. My bear register entry (post-freeze): defence-budget reversal via US continuing-resolution funding freezes, ceasefire re-pricing, or a peace-dividend drawdown over a cycle. The drawdown arithmetic (C$10.72 high, 12 Mar 2026, to the ceasefire trough) is my own, on dated closes, see the seasonality graphic in the Model section.Source: my bear register (evidence base); no single public document, named, not linked..

The fourth is the register’s highest-severity active entry, and it is my own model doing the biting. The sum of the parts says rich. Valuing each brand in the combined group on its own peer cohort blends to roughly C$5.2 per share of fair value, a level the stock sat ~51% above when the entry was logged near C$7.85, and one the 30 June 2026 close of C$6.34 still sits ~22% above; read backwards, striking the parts at peer multiples leaves the market paying ~34–40× EBITDA for the residual Kraken stub^My model · DRAFTEvidence. My sum-of-the-parts workbook values each brand (sonar, batteries, navigation, software, optical, sensing) on peer-cohort multiples and blends to ~C$5.2/share; the reverse-SOTP implies ~34–40× for the Kraken stub. Register severity: HIGH, the only active bear at that grade. All arithmetic on dated closes (C$7.85 at logging; C$6.34 on 30 Jun 2026).Caveats: cutting both waysThe workbook is DRAFT with weightings I set myself; peer multiples are vintage, struck before the underwater-drone demand ramp fully priced into comps; and SOTP structurally ignores platform and integration value, which is the entire deal thesis.Strongest contrary sourceSell-side targets sit far above the SOTP blend, the street prices the platform, not the parts: Desjardins target raise, Aug 2025 (Cantech Letter)Source: my sum-of-the-parts model (evidence base, DRAFT), my own analysis, labelled as such; no external document.. The bull answer is the thesis itself: a sum-of-the-parts prices the brands separately when the entire logic of the deal is that the integrated stack, one vendor, one balance sheet, one customer ledger, is worth more than its pieces. But that answer must be earned in integrated prints, and until it is, my standing caution applies with one addition: all five ranked failure modes above are live, several are mutually reinforcing, and the price leaves little margin for any of them to bite.^My viewEvidence. The severity ranking is my register’s grading; the mutual-reinforcement observation is mine, integration stress, cash conversion and concentration share a transmission mechanism (one program slipping moves all three); and “little margin” is priced off the 30-Jun close of C$6.34, dated.Source: my bear register (evidence base), my own grading; price dated as stated. The four bears on this card make it nine.

The bear case is not a different set of facts, it is the same facts, before they are proven.

Beyond the ranked bear case, the second-order risks: capacity execution, the Dartmouth battery plant runs ~2 years from announcement to full ramp and has already slipped a quarter; FX translation, a GBP/EUR-heavy revenue base now reports in CAD, and the deal’s own cash leg floats with GBP/CAD; pricing pressure, Kraken lost the ~C$50M Singapore MCM tender to incumbent Thales in 2025, its first acknowledged RFP loss to a larger prime, and the lone Sell on the Street builds on exactly that^Verified · loss & incumbentEvidence. The ~C$50M Singapore MCM loss (2025) is acknowledged in management’s own earnings-call Q&A (transcript in my register) and anchors the lone Sell I have logged, ATB Capital Markets, C$6.50 target (note held in my register, no public URL). Thales’s incumbency is structural, not anecdotal: it is developing modular sonar for uncrewed ASW/seabed warfare and is partnered with HII on autonomous MCM.Thales: the incumbent’s footprintNavy Lookout, Thales modular sonar for uncrewed ASW & seabed warfareHII, Thales/HII partnership on autonomous undersea MCMThe loss itselfManagement earnings-call Q&A + ATB Capital Markets Sell note, both catalogued in my register; no public deep-links logged, none composedOne lost tender is not a trend, but primes defend MCM incumbency hard, and I watch every MCM tender Kraken enters from here.Sources: independent trade press for Thales positioning · register-held items named, not linked.; and flow risk, a ~72%-retail register, thin coverage^Verified · registerEvidence. 72.4% retail / 24.3% institutional / 3.2% insider at 30-Apr-2026, my three-agent fund-ownership sweep across the public holdings tape (Fintel, N-PORT/13F filings, Jun-2026). The 2× leveraged wrapper is real and SEC-registered: Defiance “Daily Target 2X Long Kraken Robotics” ETF (ticker KRK, Tidal Trust II), filing named, no public URL logged, none composed.SourcesFintel, KRKNF institutional ownership tapeSEC Tidal Trust II registration (485APOS/497, Apr–May 2026), named from my registerName-collision hazard: Payward, the crypto exchange, also trades as “Kraken”, and daily-reset leverage decays both directions, amplifying the flow noise this paragraph counts.Sources: ownership tape dated 30-Apr-2026 · accessed Jun-2026. and a 2× leveraged ETF make the stock’s price discovery noisier than its business^Verified · mixed gradesEvidence. Dartmouth timeline from the Feb-2025 announcement vs current status (retrofit, ~C$10M, >60,000 sq ft, 200 jobs); the Singapore loss to Thales (~May 2025) is the anchor of ATB Capital Markets’ Sell thesis; ownership split per Apr-2026 snapshots. Each carries its mitigant in the base: capacity is scaling ahead of revenue; FX cuts both ways; one lost tender is not a trend, yet.Sources: Kraken, FY2025 results (facility, guidance) · analyst positioning & ownership: my register (ATB Capital Markets note; Bloomberg snapshots), catalogued in the evidence base.

Currency. Post-Covelya, a CAD-reporting Kraken earns most of its revenue in GBP, EUR, USD and DKK^Verified · brand domicilesEvidence. The currency map follows the brand domiciles: Sonardyne, Wavefront and Chelsea are UK companies (GBP), EIVA is Danish (DKK), Voyis is Canadian, and the defence order book runs heavily in USD, all consolidating into a CAD reporting line.SourcesSonardyne, Covelya Group page (the six brands)EIVA, company newsroom (Danish domicile visible first-hand)The residual is the one the prose names: once four currencies blend into one reporting line, organic growth gets harder to read from outside.Sources: company-primary pages for domiciles.. I grade the exposure two-sided and low: at roughly 1.87 GBP/CAD, logged mid-June 2026, stale by definition, translation currently works for the CAD line, not against it^My calcEvidence. My arithmetic, dated: the deal deck rate is £1 ≈ C$1.842; the spot I logged mid-Jun-2026 is ~1.87. Above deck, consolidated GBP revenue translates into more CAD, that is the tailwind, while the same spread makes the £-denominated cash leg cost more CAD until it settles. Net direction depends on timing; the label is mine, not the company’s. Refresh the rate before relying on it.SourcesKraken, Covelya acquisition announcement (deal terms & deck rate)GBP/CAD series, exchange-rates.org, in my register (no public deep-link; rate flagged stale)Sources: deal release + FX series dated mid-Jun-2026.. What stays on the ledger is narrower and real: no disclosed hedging policy, OCI swings, and four currencies blending into one reporting line that makes organic growth harder to read from outside. Capacity. The Dartmouth retrofit runs about two years from announcement to full ramp into 2027, a quarter behind its original target, scaling ahead of revenue, but the quarterly ramp commentary is the tell I watch^Verified · my re-gradeEvidence. Dartmouth: ~C$10M, >60,000 sq ft Burnside retrofit, >3× battery capacity, targeted ops Q1-2026; management on the ramp: “people moved in, equipment in; initial production this quarter, ramp steadily through the year, into 2027.” FX: GBP/CAD ~1.87 logged mid-June 2026 against the deal’s £1 ≈ C$1.842 deck rate. The two-sided/low re-grade is my judgement on that evidence, not a company framing.SourcesKraken, $34M SeaPower battery orders (Dartmouth facility details & timeline)Kraken earnings-call Q&A on the Nova Scotia ramp, management transcript in my register (user-supplied, no public link)GBP/CAD current + 5-year history, exchange-rates.org series in my register (no public deep-link; rate dated mid-Jun-2026, flagged stale)Two-sided means two-sided: a GBP slide before the cash leg settles, or after revenues consolidate, cuts against the thesis, and the company has disclosed no hedging policy either way.Sources: Kraken release + call transcript · FX series dated mid-Jun-2026, refresh before use..

Spec credibility, a flag I keep against my own bull case. Kraken’s own Chief Scientist reported measured AquaPix resolution of ~3.3 × 2.6 cm in a peer-reviewed conference paper, against the “1.9 × 2.1 cm Ultra-HD” marketing figure; the current flyer’s coverage-rate headline has softened from “4” to “up to 3.5” km²/hr^Verified · my comparisonEvidence. The softening is documented across the company’s own dated collateral in my register, the archived Sept-2025 product/website PDF against the live AquaPix product page. I compare Kraken’s documents with Kraken’s documents; nothing third-party is needed.SourcesKraken, archived Sept-2025 website/product PDFKraken, live AquaPix / SAS product page“Up to” phrasing is standard marketing hedging industry-wide; the signal here is the direction of revision, not the hedge itself.Sources: company collateral, both poles dated., and the battery energy-density “200%/3×” claim reconciles to a defensible ~2×^My calcEvidence. The reconciliation is my arithmetic across published spec tables: SeaPower’s claimed volumetric energy density set against conventional pressure-housed subsea packs, e.g. General Dynamics’ 1.5 kWh subsea battery module, lands the defensible multiple nearer 2× than the “200%/3×” headline.SourcesGD Mission Systems, 1.5 kWh subsea battery (benchmark spec)Kraken, SeaPower battery orders (claimed specs)Even ~2× in a pressure-tolerant form factor is a real advantage, the flag is about headline discipline, not product inferiority.Sources: spec tables as published; multiple is my model output.. Spec-softness is industry-wide in marine electronics, but a thesis built partly on technical superiority owes the reader the measured numbers, not the brochure ones^Verified · primary paperEvidence. The measured-vs-marketed gap comes from Kraken’s own side: its Chief Scientist’s OCEANS-2019 conference paper (IEEE) reporting measured AquaPix SAS resolution ~3.3 × 2.6 cm, set against the company’s current product-flyer claims. The energy-density reconciliation (headline “3×” → ~2× defensible) is my arithmetic across the published spec tables.SourcesDillon, OCEANS-2019 conference paper (IEEE), measured AquaPix resolution; named, not deep-linkedKraken, AquaPix / MINSAS product page & flyer (marketing specs)The honest counterweight: measured 3.3 × 2.6 cm is still class-leading against side-scan physics, and every competitor’s brochure suffers the same optimism, the flag is about credibility discipline, not product inferiority.Sources: Kraken-authored IEEE paper vs Kraken marketing collateral, the gap is documented from primary material..

Two lower-severity items complete the register. The uplisting timeline, TSX, then possibly a US listing, reaches me only through an IR-channel source: undated, from the company’s paid investor-relations firm, and gated on the 3D at Depth ERP integration^Single source · IR channelEvidence. The uplisting pathway reaches me only through Sophic Capital, Kraken’s paid investor-relations firm (emails, Oct-2025, in my register): undated, and gated on the ERP integration. The mechanism itself is standard and procedural, which is exactly why a paid-channel timeline earns no weight until a filing exists.The mechanism (verified)TMX, TSXV-to-TSX graduation guideMagna Mining, conditional graduation approval, May-2026 (the cadence a real filing follows)The timeline (soft)Sophic Capital IR emails, Oct-2025, user-supplied, no public link, graded softPathway-is-easy is not filing-is-coming: my model prices nothing here until a filing exists.Sources: exchange-primary for mechanics · IR channel named and graded.. I carry it as single-source colour, not as a catalyst my model is allowed to price. (Insider selling, the other perennial flag, is dealt with at full weight in the ownership section above, disputed magnitude, quarantined until SEDI-verified.) And deep-sea mining: the retail narrative that would make Kraken the picks-and-shovels play of a seabed-minerals boom prices a driver that stays sub-scale for years, the ISA’s mining code is still unadopted with ~40 nations backing a moratorium^VerifiedEvidence. The ISA Council session closed with the mining code still unadopted and the moratorium/precautionary-pause camp at roughly forty states, per the IISD Earth Negotiations Bulletin summary, the standing UN-adjacent record of ISA sessions.Sources: IISD ENB, ISA Council 30-2 session summary, Norway’s licensing round is paused to at least 2029^VerifiedEvidence. Norway paused its first deep-sea-mining licensing round (reported Dec-2024); subsequent reporting logged in my register extends the pause to at least 2029.Sources: CNBC, Norway pauses deep-sea-mining plans, and Kraken has zero disclosed DSM revenue^Verified · my filings passEvidence. No deep-sea-mining revenue line appears anywhere I have looked in the company’s own materials, FY2025 results, Q1-2026 results, the Mar-2026 investor deck. The absence is my read of those documents; absence from disclosure is not proof of absence of pipeline, and I treat it only as what it is.SourcesKraken, FY2025 resultsKraken, Q1-2026 resultsKraken, investor presentation, Mar-2026 (PDF)Sources: company-primary; pass is mine.. But precisely because Kraken carries no DSM-specific downside, no licences, no capex, no exposure if the industry never happens, I treat it as a free option, not a bet. And the option is not worthless: Washington has been prying the door open on its own authority^Mixed · single-source items flaggedEvidence. Uplisting: IR-firm emails (undated, gated on ERP integration), single source, paid channel, graded soft; one secondary commentary piece repeats the pathway. DSM: regulatory freeze verified across independent press and UN-adjacent reporting; the US reopening is primary (Executive Order + NOAA determination). Kraken’s zero DSM revenue is from its own filings.Uplisting (single-source, soft)Sophic Capital (Kraken’s paid IR firm) emails, Oct-2025, in my register (user-supplied, no public link)Seeking Alpha, TSXV→TSX→possible NASDAQ pathway (commentary, soft)DSM: the freezeCNBC, Norway pauses deep-sea-mining licensing (extended to ≥2029)Oceanographic, ISA 2026: mining code still unadopted, ~40 nations back a moratoriumDSM: the door reopening (contrary to the freeze)White House, Executive Order 14285, offshore critical minerals (24-Apr-2025)TMC / NOAA, deep-seabed application in “substantial compliance” (Mar-2026)The narrative risk cuts at the register, not the business: a retail base that bought a seabed-minerals story can sell it just as fast when the story stalls, that is flow risk, already counted above.Sources: primary US instruments + independent press for DSM; IR-channel items named and graded soft, no link fabricated..

The integration risk will announce itself early if it comes: watch the ERP unification under the new CIO^Company claim · in partEvidence. The CIO seat is visible on Kraken’s own leadership page; that ERP unification gates the wider integration (and the uplisting timeline) is IR-channel colour, graded soft. I choose ERP as the leading indicator because it is observable and dated when it lands, unlike “integration going well”.SourcesKraken, leadership team pageSophic Capital IR emails (ERP gating), in my register, no public link, graded softSources: company page + IR channel named and graded., brand rationalisation across the six Covelya businesses (including the two overlaps the group inherits), any disclosed cross-sell win, and, above all, whether the acquired leadership named in the retention architecture is still there in twelve months^My readEvidence. The retention architecture, management continuity across the acquired brands, is described in the deal announcement; the twelve-month test is mine: unplanned acquired-leadership exits clustered in year one are the earliest hard signal an integration is failing. Brand-level leadership in this group does turn over, EIVA’s own CEO-transition release is the on-file precedent.SourcesKraken, Covelya acquisition announcement (retention & continuity)EIVA, CEO transition announcementAn orderly, pre-planned succession is not a failure signal, the test is unplanned exits, clustered, in year one.Sources: deal release + brand newsroom; the test itself is my framework.. On the standalone book: H2 delivery against the C$165–175M guide^Company claimEvidence. The C$165–175M combined-year guide is management’s number, carried with the Q1-2026 results. The delivery history is the caveat it travels with: FY2025 was guided at C$120–135M and delivered C$102.2M, which is exactly why H2 delivery sits on my watch list rather than in my base case unexamined.SourcesKraken, Q1-2026 results (guidance)Kraken, FY2025 results (guided vs delivered)Sources: company releases; the guided-vs-delivered contrast is on the record., and whether announced intake keeps converting inside a 3–6-year window^My modelEvidence. The 3–6-year window is my estimate from a PR-timeline pass over Kraken’s dated award releases mapped against subsequent delivery disclosures. The poles on file: the ~C$50M Royal Canadian Navy minehunting award ran multi-year from award to delivered revenue, while small commercial orders, the C$3M Ocean Infinity award, sit at the short end. The window is a range, not a point.SourcesKraken, ~C$50M RCN minehunting awardKraken, C$3M Ocean Infinity awardDefence programs slip; this window is a base assumption of my revenue model, so a rightward shift hits the model directly.Sources: company award releases; window is my model output.. None of these require faith; all of them are observable.

A thesis that cannot tell you how it fails is a belief. This one publishes its failure conditions.

Every combined or forward number here is my estimate built from verified components, official combined guidance issues at close and supersedes all of it. Currency is CAD; the deal converts GBP at the £1 ≈ C$1.842 deck rate. Analysis, not a recommendation.^Verified · deck-statedEvidence. £1 ≈ C$1.842 is the deal deck’s stated conversion. Mid-Jun-2026 spot in my register ran ~1.87, so translation moves every combined C$ figure here slightly, the two-sided FX reading is treated in Risks.Source: Kraken, Covelya acquisition announcement (deal terms & deck rate)

Beneath that revenue line sits the demand ledger that feeds it. I keep a year-by-year tally of every product order Kraken has announced with a disclosed value, and the shape is unambiguous: from a single US$325k sonar contract in 2015 to ~C$97M announced by the Q1-2026 release, with the steepest rungs all in the last four years^Verified · anchor awardsEvidence. The ladder’s big rungs each trace to a primary release. 2020: the C$36M Royal Danish Navy MCM award (KATFISH towed SAS + Tentacle winch + ALARS), the year Poland also first selected KATFISH for its Kormoran II minehunters at an undisclosed value, adding nothing to the bar. 2022: the RCN Remote Minehunting & Disposal System award, announced as “C$50M+”, ~C$57.9M all-in on the federal disclosure record (C$45.8M for two systems + C$12.1M in-service support). 2025–26: the battery/SAS cadence, C$34M (Feb 2025), C$35M (Jan 2026), C$24M defence orders including a new Polish KATFISH (Mar 2026), C$28M (Apr 2026).Company primariesKraken PR, C$36M Danish Navy contract, 8 Sep 2020Kraken PR, Polish Navy minehunting systems (undisclosed value), Sep 2020Kraken PR, RCN RMDS “C$50M+”, Dec 2022Kraken PR, C$34M SeaPower orders, 26 Feb 2025Kraken PR, C$35M SeaPower sales, 13 Jan 2026Kraken PR, C$24M defence orders, 17 Mar 2026Kraken PR, C$28M SeaPower + SAS orders, 16 Apr 2026Buyer-side / independentVanguard, RMDS contracts awarded by Minister Anand (buyer-side), Dec 2022The ~C$57.9M all-in RMDS figure reads off Canada’s proactive-disclosure contract record (search.open.canada.ca, vendor: Kraken Robotic Systems Inc.), a government dataset, named here rather than deep-linked because the record sits behind a search form.Caveat“C$50M+” at announcement vs ~C$57.9M contracted shows announced values and contract ceilings drift; the 2020 and 2021 bars carry “+” because same-year undisclosed awards are invisible in this series.Sources: Kraken, Danish Navy award · Vanguard (buyer-side). The ladder is a floor, not a backlog^My register · my tallyEvidence. My year-by-year tally of every Kraken product-order announcement carrying a disclosed value, read across the company’s full press-release archive (2015–2026) and cross-checked against the annual results releases. It is a floor on true intake, not a backlog: many awards are announced with no value at all (the Polish Navy’s 2020 KATFISH selection and 8-year framework, the HII and CMRE SAS orders), and the ~C$50M/yr recurring services book is never announced as orders.Read it honestlyAnnounced flow, not audited backlog: Kraken discloses no contracted-backlog balance, so this series cannot be reconciled to any filing. Selectivity cuts both ways, so I treat the slope as the signal and each bar as a minimum. The 2015 entry is US$; all else CAD.Sources: Kraken Robotics, press-release archive (tally basis) · Kraken, FY2025 results (cross-check), 16 Apr 2026: it excludes every undisclosed award and the ~C$50M/yr recurring services book, which is why I read its slope, not its level, as the signal the model capitalises.

The margin story is the part of the trajectory I weight most, because it is the part a revenue chart cannot fake. Gross margin sat in the low-to-high-40s for six straight years, 41.7% in 2022, ~49% in 2023 and 2024, before jumping to 62.1% in 2025. My read of the jump is mix: services went from 27% to 40% of revenue with the 3D at Depth acquisition^VerifiedEvidence. The deal behind the mix shift: Kraken bought 3D at Depth for US$17M cash, announced 13 Mar 2025, closed 2 Apr 2025; 56 staff across Longmont CO, Houston and the UK; 2024 revenue US$14M at a ~60% gross margin, ~20% three-year CAGR; rebranded as Kraken on 30 Sep 2025.Sources: Kraken, acquisition announcement, 13 Mar 2025 · completion release, 2 Apr 2025 · rebrand, 30 Sep 2025, and the product book got richer, batteries and sonar systems rather than lower-margin integration work. Kraken does not disclose segment-level gross margins, so that attribution is my inference from the disclosed mix shift, not a company-stated split^Verified · attribution is my readEvidence. Gross margin 41.7% (2022), 48.8% (2023), 49.0% (2024), 62.1% (2025), all from audited results; services 27% → 40% of revenue across the 3D at Depth year. The causal attribution of the 2025 jump to mix is my inference, flagged as such because the company publishes no segment margins to confirm it.SourcesKraken, FY2025 results (62.1% gross margin, mix disclosure)FY2021–FY2024 results releases and Q4-2025 MD&A, SEDAR+ filings, reconciled in my registerA one-year margin jump can also carry contract-timing and revenue-recognition effects; a second year at this level is the confirmation my model waits for.Sources: Kraken audited results releases; segment-mix attribution labelled as my inference..

Now the caveat, the one my own moderator discipline insists I print next to the ladder rather than under it. The 2025 Adjusted-EBITDA figure of C$25.0M was lifted mainly by adding back about C$5.0M of acquisition and restructuring costs. Strip those out and unadjusted EBITDA actually slipped slightly year-on-year, roughly C$20.9M to C$20.0M, as headcount and overhead grew with the company.^My calcEvidence. FY2025 Adjusted EBITDA of C$25.0M includes roughly C$5.0M of acquisition and restructuring add-backs; strip them and unadjusted EBITDA ran ~C$20.9M (2024) → ~C$20.0M (2025), my arithmetic off the audited statements and the company’s own reconciliation. Both readings printed: the add-backs are genuinely deal-driven (Covelya, 3D at Depth) and arguably one-off; Adjusted is the company’s chosen metric, unadjusted is the stricter one.Sources: Kraken, FY2025 results · Q4/FY2025 call transcript (Investing.com) · my calc. Not investment advice. Reported net income is an even poorer guide across these two years: 2024’s C$20.1M was flattered by a one-off C$9.7M deferred-tax benefit, and 2025’s C$2.9M was depressed by the same C$5.0M of deal costs plus a normalised tax charge. So the honest lens is Adjusted EBITDA read with the add-back visible, which is exactly how the chart above renders it. The step-up survives the audit; it just arrives one notch lower than the headline^Verified · my ledgerEvidence. 2025 Adj EBITDA C$25.0M includes ~C$5.0M of acquisition/restructuring add-backs; unadjusted EBITDA ≈C$20.9M (2024) → ≈C$20.0M (2025). Net income: C$20.1M (2024, incl. one-off C$9.7M deferred-tax benefit) vs C$2.9M (2025, after deal costs and a normalised tax charge). All reconciled from the results releases, MD&A and audited statements in my financial-anatomy ledger.SourcesKraken, FY2025 results (Adj-EBITDA definition and add-backs)Q4-2025 MD&A + audited FY2025 statements, SEDAR+ (named, not deep-linked)The strongest contrary reading: if add-backs recur post-Covelya (integration is expensive), “adjusted” and “economic” EBITDA keep diverging, my model watches the gap, not just the headline.Sources: Kraken audited filings; the unadjusted bridge is my arithmetic on disclosed figures..

Margins are only half the earnings-quality file; the cash line is the other half, and in 2025 it is the harder reading. Kraken turned C$25.0M of Adjusted EBITDA into just C$1.5M of operating cash, a conversion rate of roughly 6%, down from about 56% in 2024, because scaling receivables and inventory absorbed an estimated C$18–20M of working capital.^My calcEvidence. Operating cash of C$1.5M against C$25.0M Adjusted EBITDA ≈ 6% conversion; the same ratio ran ~56% in 2024, my ratios off the audited cash-flow statements. The C$18–20M working-capital absorption is my estimate from the receivables and inventory movements; the C$30.3M of capex and capitalised development is as reported. Caveat at full weight: single-year conversion ratios are noisy in a scaling hardware business, the direction is the finding, not the decimal.Sources: Kraken, FY2025 results (cash-flow statement) · my calc. On top of that sat C$30.3M of capex and capitalised intangibles, 29.6% of revenue, against a 2022–2024 range of ~6–12.5%, driven chiefly by the new Nova Scotia battery plant. Free cash flow landed around −C$29M, and the build was funded by equity, chiefly the C$115M bought deal of July 2025, not by operations^Verified · filings + my bridgeEvidence. FY2025: operating cash flow C$1.5M against C$25.0M Adjusted EBITDA (≈6% conversion), versus C$11.6M against C$20.7M in 2024 (≈56%); capex + capitalised intangibles C$30.3M = 29.6% of revenue (2024: ~C$5.1–5.8M, ≈6%). Free cash flow ≈−C$28.8M is my arithmetic (CFO minus capex); the working-capital bridge, C$18–20M absorbed, receivables −C$10.9M and inventory −C$10.0M, is my estimate from the cash-flow statement, flagged as such. All figures reconciled in my financial-anatomy ledger: a four-agent pass I ran across the FY2021–FY2025 results releases, the Q4-2025 MD&A and audited statements, and the Q1-2026 interim, tying revenue mix, margins, capex, working capital and cash flow into one dated series.Primary filingsKraken, FY2025 results (Adj EBITDA C$25.0M; capex disclosure)Kraken, Q4 & FY2025 MD&A, 15-Apr-2026 (cash flow, working capital, capex detail)July-2025 C$115M bought-deal financing, company release logged in my register (named, not deep-linked)The bull rebuttal is real and carried here at full weight: defence hardware consumes cash ahead of revenue, and the plant is capacity growth, not maintenance spend. It does not change the 2025 fact pattern: operations did not fund the year, shareholders did.Sources: Kraken FY2025 results · Q4-2025 MD&A (SEDAR+) · FCF and working-capital bridge are my arithmetic/estimate on those filings..

I read those two panels the way the bear case does before I answer them. The honest answer has three parts. First, the spend is capacity, not decay: the Dartmouth battery plant is what lets SeaPower scale^Verified · specs company-statedEvidence. The Nova Scotia battery build is documented in Kraken’s own SeaPower orders release: a ~$10M retrofit of >60,000 sq ft in Dartmouth’s Burnside Industrial Park, adding more than 3× battery capacity, operational Q1-2026. That is what “capacity, not decay” rests on. Caveat: the capacity multiple and the timeline are the company’s stated figures, not independently measured.Source: Kraken, $34M SeaPower battery orders (Dartmouth capacity detail), and growth in defence hardware consumes receivables and inventory ahead of revenue, the pattern is normal for the phase. Second, the funding was clean: equity raised ahead of need, not debt strain. But third, and this is the row my model refuses to soften, a franchise that converted ~56% of EBITDA to cash in 2024 converted ~6% in 2025, and the return to self-funding is a 2026 thesis to prove, not a 2025 result to bank. That is why my ledger leaves the 2026E conversion cell deliberately blank: I expect capex intensity to normalise toward ~9–11% as the plant completes, and I will mark conversion when the cash-flow statement prints it, not before^My model · normalisationEvidence. 2026E capex C$15–18M (≈9–11% of guided revenue) is my estimate from the completed plant build and management’s capex commentary; 2026E cash conversion is left blank in my ledger on purpose, estimating it would launder the open question the bear case turns on. What moves this cell: the first combined reporting cycle showing operating cash flow tracking Adjusted EBITDA through the H2-weighted ramp.BasisMy financial-anatomy ledger, the four-agent filings pass described above; normalisation assumption flagged as my estimateKraken, Q1-2026 results (guidance reiterated; the year the cell waits on)The contrary case, full weight: FY2026 guides ~+65% revenue growth, and growth at that rate rebuilds receivables and inventory, conversion can stay pinned low for a second year for entirely mechanical reasons, and a second equity-funded year would start looking like a pattern, not a phase.Sources: my model (assumptions stated) · Kraken Q1-2026 results..

The mix behind the margin, from the calls^Their words · call transcriptEvidence. Q4-FY2025 call (16 Apr 2026): product-line growth (battery +~30 / SAS +~60 / services +60 percent; 2025 SAS units ≈ prior two years combined); 2026 GM guide 55–60 percent with the one-time Q4 low-cost item flagged by management itself. Q4-FY2024 call: services organic >15 percent (>20 incl. 3D at Depth); US offshore-wind excluded from the forecast. Company claims on recorded calls.Sources: Q4/FY2025 call (transcript) · Q4/FY2024 call (transcript): FY2025 battery revenue up ~30 percent, SAS up ~60, services up 60, with 2025 SAS unit shipments roughly equal to the prior two years combined, a 2026 gross-margin guide of 55–60 percent (management itself flagging that Q4’s print was helped by a one-time low-cost item), services growing “north of 15 percent” organically, and no US offshore-wind assumed in the forecast at all, conservatism you can quote.

The model is built off the pro-forma structure, not the pre-deal one: 307.2M shares + 47.4M subscription receipts + 15.9M consideration shares ≈ ~370–385M fully diluted, with pro-forma net debt of only ~C$26–100M depending on the measurement date, leverage of ~0.3–0.8×, close to debt-free for a C$615M acquisition^My reconstructionEvidence. Components verified (MD&A share count 307,175,048; 47,353,550 receipts; 15,882,352 consideration shares per the SPA; C$150M facility undrawn); the assembled pro-forma is my arithmetic. The model’s C$100M debt / ~372M share basis is deliberately the conservative corner.Sources: Kraken Q4-2025 MD&A (15 Apr 2026) · SPA, SEDAR+ 06406273 · Kraken, receipt offering close. On earnings, the honest choice is stated rather than hidden: on a full run-rate basis combined 2026E revenue is ~C$455–480M (mid ~C$465M, Kraken’s C$165–175M guide plus a ~C$295M Covelya estimate)^My modelEvidence. Mid ~C$465M = the C$170M midpoint of management’s C$165–175M standalone guide (company-stated, forward-looking) plus my ~C$295M Covelya 2026E, my one-year-forward extension of the prospectus supplement’s series (target revenue C$169.5M → 214.0 → 249.0–275.2 in 2025E) at the filing’s own trajectory. Official combined guidance issues at close and supersedes this assembly.SourcesKraken, Q1-2026 results (guide reiterated)Kraken investor presentation, Mar-2026 (PDF)The Covelya leg is mine, not the company’s: the prospectus series stops at 2025E, so my 2026E is extrapolation, and the partial-year consolidation optic is materially smaller than run-rate.Source: my valuation model. Estimate; not investment advice.; the partial-year optic (~mid-2026 close) is materially smaller. Multiples quoted below use run-rate.

Dilution is the bear point I can draw most precisely, so here it is drawn^Verified · my waterfallEvidence. My fully-diluted waterfall, built line-by-line from the primary record (16 Jun 2026): basic 307,175,048 (Q4-2025 MD&A); +47,353,550 subscription receipts converting 1:1 at close, prospectus-qualified so free-trading immediately; +15,882,352 consideration shares fixed at C$8.50 by SPA clause 5.1; +13,715,750 options (WAEP C$1.313, ~5.75M exercisable now, expiries to Oct-2032). No warrants; no RSU/DSU plan found (absence single-sourced). Ties management’s stated ~385.1M FD within ~0.3%.Primary sourcesKraken, C$402.5M subscription-receipt offering close, 12 Mar 2026Kraken, acquisition announcement (lock-up terms), 3 Mar 2026 · Q4-2025 MD&A + Share Purchase Agreement clause 5.1 (SEDAR+ filings, in my evidence base)The honest overhang notesA 25-month base shelf (filed 7 Aug 2025, good to ~Sep 2027) is standing, uncapped issuance capacity; and I found no insider/management deal lock-up in the public materials, an absence, not a confirmed freedom, but worth knowing., every share that exists or can exist, and exactly when each tranche can trade:

The read: the +25.1% basic-to-FD step is real but fully mapped, the only unannounced overhang is the standing shelf^VerifiedEvidence. The shelf is real and dated: Kraken announced the filing of a base shelf prospectus on 7 Aug 2025, standing capacity to issue securities without a fresh prospectus, and the one dilution lever the March raise did not map to a schedule.Source: Kraken, base shelf prospectus filing, 7 Aug 2025. The seller’s 15.9M shares cannot touch the market for a year and then arrive in thirds; the receipts were absorbed at C$8.50 in March by choice. Dilution here is a schedule, not a surprise.

Kraken’s standalone anatomy nets to an improving, higher-margin franchise with three soft spots I will not soften: 2025’s growth was bought, not grown; the revenue line leans on one inferred customer; and expansion still consumes cash^Verified · in partEvidence. The FY2025 statements show the absorption directly (the conversion arithmetic is carded above); the pattern was visible mid-year too, the Q2-2025 call print was revenue growth with an EPS miss. Caveat: an EPS miss is an accrual reading, not a cash one, the load-bearing evidence is the audited cash-flow line; the transcript is a timeline marker.Sources: Kraken, FY2025 results · Q2-2025 call transcript (Investing.com). The deal is the designed cure for all three, and until the 5 March 2026 prospectus supplement, the thing being bought was known mostly second-hand. That filing carries Covelya’s audited FY2023 and FY2024 consolidated statements, the first primary disclosure of the target’s standalone economics, and it reframes the deal^Verified · primary filingEvidence. Prospectus Supplement dated 5 March 2026 (to the August 2025 base shelf), carrying Covelya’s audited consolidated FY2023/FY2024 statements and Kraken’s pro-forma statements in its schedules; the target and combined series below read straight off its summary tables. All figures CAD at the filing’s stated deck (£1 = C$1.7504 in 2024; C$1.8420 in 2025).Sources: Prospectus Supplement, 5 Mar 2026 (SEDAR+, filed under Kraken Robotics Inc.) · user-supplied filing in the evidence base. On audited FY2024 numbers Covelya generated C$214.0M of revenue and C$50.6M of Adjusted EBITDA against Kraken’s C$91.3M and C$20.7M, the target is roughly 2.3× the acquirer’s revenue and 2.4× its EBITDA, at a near-identical margin (≈23.6% vs ≈22.7%), with Adjusted EBITDA up ~59% in a year. A reverse-scale acquisition of a comparable-margin business growing this fast is not a bolt-on; it is the smaller company buying one more than twice its size, which is simultaneously the boldest fact in the file and the reason integration tops the bear register.

The series the model stands on, straight off the filing: target revenue C$169.5M → 214.0 → 249.0–275.2 (2025E); combined revenue C$239.1M → 305.3 → 351.0–379.2 (2025E); combined Adjusted EBITDA C$45.9M → 71.3 → 84.3–92.7 (2025E), margins walking 19% → 23% → 24%^Company figures · illustrativeEvidence. Management-reported combination by simple aggregation. The filing’s own caution, carried at full weight: this “Combined Financial Information” is illustrative only, not IFRS-3 pro-forma, excluding purchase-price allocation, fair-value step-ups, transaction costs, financing effects and synergies; the 2025E component is preliminary, unaudited and management-estimated.Source: Prospectus Supplement, 5 Mar 2026 (SEDAR+). Three honesty flags travel with those numbers wherever they appear on this site. Covelya’s 2023 Adjusted EBITDA leans on a C$4.6M non-recurring legal add-back, without it the 2023 base is C$26.2M^Verified · registryEvidence. The statutory floor under the illustrative series: Covelya Group FY2024 accounts (Companies House 12493148), revenue ~£122.3M, net assets £86.3M, cash £31M, 617 employees; Sonardyne International standalone £78.7M turnover; Sonardyne’s FY2025 full accounts filed 2 Apr 2026. The C$4.6M add-back and the C$26.2M ex-add-back 2023 base are the prospectus’s own footnotes (named, no deep link).Registry (primary)Companies House, Covelya Group Ltd (12493148)Companies House, Sonardyne International (01299452), filing historyCorroborationMarine Technology News, two 2024 King’s Awards (Sustainable Development + International Trade)Statutory GBP figures and the filing’s CAD illustrative series differ by FX and basis; the add-backs are management-defined., so part of the 2023→2024 growth optics is a depressed base, and the add-backs are management-defined. The combined figures exclude everything IFRS will not: reported net income after close will sit well below what combined Adjusted EBITDA implies once amortisation of stepped-up intangibles and new interest expense land. And 2025E is an estimate wearing an audited filing’s clothes, preliminary, unaudited, management’s.

The same filing pins the capital architecture the scenarios run on: 363,451,441 shares post-offering, 369,627,991 with the over-allotment, which was taken up in full at the 12 March 2026 closing, hardening the ~372M count used above^Verified · filing + closeEvidence. Capitalization table: 363,451,441 shares post-offering, 369,627,991 with over-allotment; over-allotment exercised in full at the 12 Mar 2026 closing (C$402.5M gross at C$8.50). Net cash of C$(88.5)M at 30 Sep 2025; adding the C$150M facility and Covelya leases gives Combined Net Debt ~C$70.2M ≈ 0.8× midpoint 2025E combined Adjusted EBITDA, a denominator that is itself the illustrative EBITDA the filing cautions on. The same-price-as-consideration read is my inference.Sources: Prospectus Supplement, 5 Mar 2026 (SEDAR+) · Kraken, closing of C$402.5M subscription-receipt offering. Kraken walked into a C$615M acquisition in a net-cash position; the combined book carries ~0.8× leverage, the conservative end of near-debt-free. The equity was raised as subscription receipts held in escrow at C$8.50, refundable if the deal broke, so dilution crystallised only at close, and C$8.50 is the same price Sonardyne’s owners took their consideration shares at. The vendor took stock at the marketed price, not a premium: the sellers who know the asset best chose paper over cash at the deal price^VerifiedEvidence. The consideration record: C$615M total, $480M cash (bought deal + term facility) plus $135M in Kraken shares; 47.3M subscription receipts at C$8.50, closed 12 Mar 2026; signed 3 Mar 2026.Deal record (primary)Gowling WLG (Kraken’s deal counsel), client-work record of the C$615M structureIndependent corroborationBetaKit (Alex Riehl, 4 Mar 2026), $480M cash + $135M shares; ~1,200 employees / 450,000 sq ft post-closeWhether stock-at-the-deal-price signals seller conviction is my inference, consideration mix is also a tax and negotiation artefact.. That is my inference, and it is the right kind of tell.

The fine print of the financing architecture is itself a risk-mitigant, and worth reading closely. The starting point at 30 September 2025 was 306,392,088 shares (plus 13.7M options), C$232.1M of share capital and no drawn acquisition facility; the build reconciles line by line, 306.39M, plus 41.18M subscription receipts, plus 15.88M consideration shares, equals 363.45M; add the 6.18M over-allotment and you land on 369,627,991, with share capital stepping C$232.1M → C$763.3M and shareholders’ equity to C$758.6M^Verified · prospectus + closeEvidence. Capitalization tables, facility terms and use-of-proceeds from the 5 Mar 2026 prospectus supplement; over-allotment take-up per the 12 Mar 2026 closing release. The ~7%-discount observation is my arithmetic on two dated prints (C$8.50 issue vs C$9.15 close, 4 Mar 2026, the last session before filing). The ~0.8× leverage denominator remains the illustrative combined EBITDA the filing itself cautions on.Sources: Prospectus Supplement, 5 Mar 2026 (SEDAR+) · Kraken, C$402.5M offering close, 12 Mar 2026.. The debt leg is a C$150M Bank of Nova Scotia term facility, committed, secured, maturing 3 March 2031, guaranteed by Kraken and the material Covelya subsidiaries, priced at a margin over Canadian prime that ratchets with the funded-debt/EBITDA ratio, sitting alongside an existing C$35M revolver, a C$10M capex line, a C$10M letter-of-credit line and a C$30M accordion^Verified · primary filingEvidence. Facility terms from the filing: C$150M committed, secured term facility maturing 3 Mar 2031, guaranteed by Kraken and the material Covelya subsidiaries, priced over Canadian prime on a funded-debt/EBITDA ratchet, beside the C$35M revolver, C$10M capex line, C$10M letter-of-credit line and C$30M accordion. The receipts offering funding the cash leg was co-led by Scotiabank and Desjardins.Sources: Prospectus supplement (Mar 2026), credit-facility notes, SEDAR+ filing I read directly; named, no public deep link · Gowling WLG deal record · Kraken, receipts closing, 12 Mar 2026. The equity leg raised net ~C$336M base (~C$386M with the over-allotment taken) as escrowed subscription receipts, refundable at the C$8.50 issue price plus interest had the deal failed by the 31 December 2026 long-stop, so the dilution crystallised only when the deal did. And the C$8.50 print was struck at only a ~7% discount to the C$9.15 close of 4 March 2026, the last session before filing: a C$402.5M raise done that tight to market, in a name this size, is its own evidence of institutional demand.

That warning has numbers, and the same filing prints them, buried in its schedules: the most useful expectation-setter in the whole filing. Schedule F-3 is the prospectus’s actual IFRS-3 pro-forma, the statements that do include the purchase-price allocation, the fair-value step-ups, the transaction costs and the new interest that the headline “combined” tables exclude, and I treat it as the honest guide to what reported GAAP earnings will look like after close. Run FY2024 as if the deal had completed on 1 January 2024 and two profitable companies become a paper loss: Kraken earned C$20.1M and Covelya C$28.7M, C$48.7M between them, yet the pro-forma consolidated line is a net loss of C$38.5M, minus thirteen cents a share^Verified · filing pro-formaEvidence. Schedule F-3, unaudited pro-forma consolidated statement of profit (loss), FY2024, as if the acquisition and financings had completed on 1 January 2024 (CAD, at the filing’s GBP deck): Kraken net income C$20.1M (audited) + Covelya C$28.7M (audited) = C$48.7M simple; pro-forma consolidated net loss C$(38.5)M, basic EPS −C$0.13 on 288.7M weighted shares, the same year whose “combined” Adjusted EBITDA reads C$71.3M. As-if FY2024 pro-forma revenue C$305.4M; loss from operating activities C$(42.0)M. The pre-merger column is pure presentation reclassification and nets to zero on the bottom line; every dollar of the swing is the post-merger deal accounting itemised in the ledger below.Why the two disclosures disagreeThe summary “Combined Financial Information” (carded above) is simple aggregation, explicitly not IFRS-3 pro-forma; Schedule F-3 is the IFRS-3 version, with purchase-price allocation, fair-value step-ups, transaction costs and financing effects included and synergies excluded.The filing’s own full-weight caution, carried here: the pro-forma adjustments are preliminary, made for illustration, may differ materially when recorded, and “undue reliance should not be placed” on these statements.Source: Prospectus Supplement, 5 Mar 2026 (SEDAR+, Schedule F-3, Kraken Pro Forma Financial Statements); user-supplied filing in the evidence base.. The bridge from one to the other is the anatomy of what purchase accounting does to this deal:

The read that matters is which of those charges survive year one. Of the C$106.2M of pre-tax post-merger charges, roughly C$75.2M is one-time or time-limited, the inventory step-up burns off as the acquired stock sells through, the transaction costs happen once, and the retention bonuses expire at the 12- and 24-month anniversaries, leaving a structural drag of about C$31M a year pre-tax: ~C$22M of acquired-intangible amortisation plus ~C$9M of facility interest, partly shielded by deferred tax^My calcEvidence. My split of Schedule F-3’s C$106.2M of pre-tax post-merger FY2024 charges into one-time versus structural, using the filing’s own notes: one-time or time-limited ≈ C$75.2M (inventory step-up C$47.9M, recognised through cost of sales over ~12 months as acquired stock sells through; transaction costs C$8.4M; retention bonuses C$19.0M expiring at the 12- and 24-month anniversaries); recurring ≈ C$31.0M pre-tax (C$22.2M incremental amortisation of stepped-up intangibles and PP&E per note 3(c), C$8.8M facility interest per note 3(f)), against which the FY2024 statement books a C$18.9M deferred-tax recovery.Caveats on the recurring figureInterest assumes the C$150M term facility fully drawn at Term CORRA-based pricing; draw less, pay less.The amortisation line rests on a preliminary purchase-price allocation, the C$204M intangibles figure and its useful lives will be finalised only after close, and the filing says the final figures could be materially different. My ~C$31M/yr is a deck-vintage estimate, not a forecast.Source: my classification of Schedule F-3 notes 3(c)–3(h), Prospectus Supplement, 5 Mar 2026 (SEDAR+). Estimate; not investment advice.. The filing’s cleaner preview is its nine-month 2025 pro-forma, where that step-up has already burned off: revenue C$268.4M, pro-forma net income +C$10.3M, C$0.03 basic^Verified · filing pro-formaEvidence. Schedule F-3, nine months ended 30 September 2025 (same as-if-1-Jan-2024 basis, so the FY2024 statement has already absorbed the one-time inventory step-up): pro-forma revenue C$268.4M; pro-forma net income C$10.3M, basic EPS C$0.03 on 337.7M weighted shares. The period still carries the structural items, C$14.5M incremental amortisation, C$6.0M facility interest, C$4.7M retention bonuses, net of a C$5.3M deferred-tax recovery. Kraken standalone earned C$2.8M and Covelya C$27.4M in the same nine months; the C$(19.9)M gap is the recurring deal accounting at work.Source: Prospectus Supplement, 5 Mar 2026 (SEDAR+, Schedule F-3, Kraken Pro Forma Financial Statements); user-supplied filing in the evidence base., profitable, but a fraction of what the Adjusted-EBITDA optics imply. That gap is not a scandal; it is arithmetic. But it is exactly the arithmetic the first consolidated prints will wear, and I would rather price it here than meet it in an earnings headline. When that headline arrives, on schedule, F-3 is the page I will be holding.

The same schedule shows what the balance sheet becomes: C$1,079M of total assets carrying C$200.2M of goodwill and C$204.0M of acquired intangibles, Technology, Customer Relationships, Backlog and Brand, plus a C$67.1M step-up in deferred tax liabilities, against C$561.4M of net assets acquired^Verified · preliminary PPAEvidence. Schedule F-3 pro-forma statement of financial position (as at 30 Sep 2025): total assets C$1,079M; goodwill C$200.2M; other intangible assets C$204.0M (Technology, Customer Relationships, Backlog and Brand, income-based and relief-from-royalty valuations); inventory fair-value step-up +C$47.9M; freehold-property step-up ~C$16.9M (third-party appraisals); deferred tax liabilities up C$67.1M on the temporary differences those step-ups create. Net assets acquired C$561.4M = C$416.2M cash + C$145.2M share consideration.Vintages: where F-3 and the executed deal differF-3 models the financing as marketed: 44.28M subscription receipts at C$7.91 (~C$350M) and 17.08M consideration shares priced off the 27-Feb-2026 close. Actuals came in tighter and larger: 47.35M receipts at C$8.50 (C$402.5M gross, over-allotment in full, closed 12-Mar-2026) and SPA-fixed consideration of 15,882,352 shares (C$135M at C$8.50), the counts my capital-base numbers carry. Kraken, closing of C$402.5M subscription-receipt offeringThe purchase-price allocation is preliminary and will be finalised after close; the cash consideration is subject to post-closing adjustment under the SPA. Both can move the goodwill/intangibles split, and with it the amortisation drag above.Source: Prospectus Supplement, 5 Mar 2026 (SEDAR+, Schedule F-3, Kraken Pro Forma Financial Statements); user-supplied filing in the evidence base.. Well over a third of the pro-forma balance sheet is deal paper. While the businesses perform, that is an accounting curiosity; if integration disappoints, it is an impairment surface, one more reason integration tops my bear register rather than footnoting it.

At the 30-Jun-2026 close of C$6.34 the combined vehicle carries an EV of ~C$2.4–2.5B: ~5.3× run-rate revenue, ~20× run-rate EBITDA (the mid-June model snapshot at ~C$7.12 read ~6.0× and ~23×; the card keeps both bases)^My calcEvidence. EV ≈ market cap at the 30-Jun-2026 close of C$6.34 on ~372–385M fully diluted shares, plus the model’s net-debt corner ≈ C$2.4–2.5B; ~5.3× my ~C$465M run-rate revenue, ~20× my run-rate EBITDA band. The mid-June model snapshot (~C$7.12) read C$2.77–2.81B, ~6.0× and ~23×; both bases dated.Sources: StockAnalysis, PNG quote & statistics (accessed Jun-2026) · my valuation model.. That is a modest premium to the true subsea pure-plays on revenue (Exail ~4.6×; Kongsberg ~4.1×) and in line on EBITDA (Kongsberg ~26×, Teledyne ~20×), still a large premium to naval primes (~13–16× EBITDA), and far below the 14–60× defence-tech story cohort^My calc · datedEvidence. Multiples computed 10–12 Jun 2026 from delayed market data and peer filings; the partial justification is fundamental: best-in-class ~24–25% EBITDA margin vs HII 9% / GD 12% / Thales 14% / Saab 11–15%. What it means: the market already pays a defence-tech-growth premium, the re-rate case must be earned by delivery, not discovery.Sources: my comp table; price StockAnalysis / TMX (delayed) · peer multiples Capstone ALSS M&A update. The strongest single valuation argument is the deal itself: paying ~9.7–10× FY25E EBITDA for a 24%-margin asset growing EBITDA ~41%/yr, into a vehicle trading at ~20× at the 30-Jun close, roughly two turns below the defence-M&A average and well below the A&D median.

Pulling the cohorts apart is where the debate gets honest. The closest subsea pure-plays, Exail at ~4.6× revenue and ~17–22× FY26E EBITDA (the purest comp, and the one whose EBITDA the data providers dispute)^Verified · dispute flag is mineEvidence. Exail’s own FY2025 presentation is the primary read: order intake +87% to €844M (including the ~€400M Q1 MCM order), core operating profit ~€103M, a 2026 double-digit-growth / ~25%-EBITDA frame. The “disputed” flag is mine: across the aggregator pages in my comps pass the FY26E EBITDA multiple spans ~17–22× because providers disagree on what counts as Exail’s EBITDA (core vs reported).SourcesExail Technologies, FY2025 results presentation (PDF)exail-technologies.com (accessed 14-Jun-2026)The purest-comp label cuts both ways: Exail’s surging MCM order book is precisely the demand my bear case says primes and rivals can capture first.Multiple readings: my comps pass (methodology in the bear-rung card)., Kongsberg at ~4.1× and 26.2×, Teledyne at ~5.0× and 20.2×, cluster near ~4.3× revenue^My comps passEvidence. My three-agent comparable-valuation pass: live multiples for the subsea cluster, defence-tech cohort and naval primes pulled from stockanalysis.com, valueinvesting.io and MarketScreener statistics pages, accessed 12–16 Jun 2026, mostly single-provider readings, so each is indicative. On revenue the cluster is tight: Exail ~4.6×, Kongsberg ~4.1×, Teledyne ~5.0×, ~4.3× centre.CaveatThe EBITDA readings do not cluster (~17–26× across the same three names), “in line with the pure-plays” is a revenue-multiple statement, and I say so.Source: my comps pass (methodology above); provider pages logged in my register.: against the names most like it, Kraken trades in line to a modest premium. The naval and diversified primes, Huntington Ingalls 12.7×, General Dynamics 15.7×, Thales 14.9×, Babcock 10.5×, Fincantieri 10.3×, sit at a ~13.5× EV/EBITDA median and ~1.5× revenue: anchor there and Kraken looks expensive, though those multiples price whole conglomerates whose subsea content is a sliver. The defence-tech / autonomy cohort is bimodal: a mostly pre-profit story tier (Palantir ~60× revenue; Ondas; Red Cat; private Anduril at ~14–28× revenue on its US$61B May-2026 post-money) against a fundamentals-anchored profitable tier (Leonardo DRS, AeroVironment, Kratos at ~3.5–6.7× revenue), against the story tier Kraken is outright cheap. And precedent transactions cluster at ~2.3–3.1× revenue and ~10–15× EBITDA.^My registerMy register. The cluster is my deal-tape log: one row per subsea / defence-sensor precedent, price and revenue/EBITDA basis taken from the acquirer’s own release or filing where disclosed.Tape points (disclosed)Teledyne – Valeport + Adimec, US$123.6M aggregate (SEC 8-K, Q2-2024)Karman – Seemann Composites + Material Sciences, US$220M ($210M cash), Jan 2026Strongest contraryHII – Hydroid, US$350M (2020, acquirer’s close release) at ~24× EBITDA per the coverage, the tape’s outlier sits far above my 10–15× bandUSNI News, HII CEO on the $350M Hydroid deal (4 Feb 2020)CaveatMost precedent multiples are undisclosed; the range rests on the disclosed subset plus broker tape, small sample, wide error bars. What partially justifies the premium to the fundamentals cohorts is quality, not narrative: a best-in-class ~24–25% EBITDA margin against HII’s 9%, GD’s 12%, Thales’s 14%, Saab’s 11–15% and AeroVironment’s 9%, stacked on above-peer growth and a deepening defence mix. So the re-rating debate is explicit, and both sides are on this page: bears anchor on subsea pure-plays, primes and the deal tape (~10–17×) and call it expensive; bulls anchor on defence-tech growth and the Anduril line and call it cheap. My view sits with the bulls for a reason the reader can check against the dates: since the mid-June snapshot the multiple has compressed toward the bear anchor at precisely the moment the approval risk died and the growth case hardened, and the mechanism that would entrench the premium, TSX graduation and index inclusion, is dated on the catalyst ladder^My viewEvidence. The cohort you anchor on is the valuation, so I state my anchor rather than pretend neutrality. I take the growth anchor for dated, falsifiable reasons: approvals cleared 18 Jun, close expected 2 Jul, guidance intact, and the price walked from the C$7.12 snapshot to C$6.34 while all of that hardened. If growth or margin misses, the prime/deal-tape anchor governs and the ~C$5.2 sum-of-the-parts bear is the fair pushback. First person marks ownership, not confidence: this is inference, not a sourced fact.Source: my assessment over the comp table, the approval record and the catalyst ladder..

The strongest valuation fact in the file is not a peer screen, it is the acquisition itself. Kraken is paying ~2.2–2.5× FY25E revenue and ~9.7–10× FY25E Adjusted EBITDA (≈12× FY2024 audited) for Covelya: a profitable, ~24%-margin asset compounding Adjusted EBITDA at ~41%/year^Verified · deal termsEvidence. Consideration ~C$615M = C$480M cash + C$135M in 15,882,352 shares at C$8.50, per the executed SPA and offering documents; Covelya FY2024 audited: C$214.0M revenue / C$50.6M Adjusted EBITDA. The ~9.7–10× is struck on FY25E forecast EBITDA (a management-derived denominator); ~12× on FY2024 audited. The ~41%/yr Adjusted-EBITDA growth is the filing’s own series.PrimaryKraken, C$615M Covelya acquisition release, 3 Mar 2026Prospectus supplement, 5 Mar 2026, SEDAR+ (access via sedarplus.ca).Benchmarks · B-grade league tablesCapstone, ALSS 2025 average 14.0× EV/EBITDA / 2.4× EV/revenue (PY 14.7×/2.7×)PCE, A&D LTM-to-Q1-2026 median 18.87× TEV/EBITDA / 3.74× TEV/revenuePrecedent deal documentsSEC 8-Ks (ESCO; HII→Alion; Teledyne→FLIR) · Naval News / CFI Group (iXblue→Exail) · Capstone / SpaceNews (Redwire→Edge Autonomy), named; the per-deal links live in my register, not composed here.Several precedent EBITDA multiples were never disclosed and rest on advisor estimates; the Covelya multiple depends on a forecast denominator. Terms are primary-verified; the comparison layer is B-grade.Sources: deal terms primary (release + SEDAR+ filing); precedent benchmarks as linked above.. Set that against the deal tape and the price stands out: roughly two turns below the Capstone ALSS defence-M&A 2025 average of 14.0× (14.7× the year before), well below the PCE aerospace-and-defence median of 18.87× across 137 LTM deals, a tape whose revenue multiples were the richest in a four-year lookback, and below every close comparable: the near-twin Redwire→Edge Autonomy at 12.9× EBITDA and 4.2× revenue (January 2025), the closest subsea precedent iXblue→Exail at ~15× (2022), and HII→Alion at 12.2×.

Why did a scarce, profitable defence-subsea asset go at the cheap end of the tape? The record suggests structure, not defect: a private-equity seller with no visible competitive auction, a commercial-subsea revenue mix that deal models price below pure defence, a GBP cross-border transaction, and 22% of the consideration taken in stock^Verified · ratio is my calcEvidence. Consideration mix per the definitive agreement: cash plus 15,882,352 consideration shares. The ~22% is my ratio of the stock leg valued at the deal price against the ~C$615M total. One honesty line: “no visible competitive auction” is inference from the absence of process disclosure, not a confirmed process detail.Sources: Kraken, Covelya acquisition announcement · share purchase agreement, SEDAR+ filing 06406273 (no public deep-link). Ratio: my calc.. My verdict on the precedent set is cheap-to-fair, and the seller taking paper at the deal price is the tell that fair is the right end of the range. The arbitrage then writes itself: EBITDA bought at ~10–12× lands inside a vehicle the market values at ~20× run-rate (30-Jun close), so every acquired dollar is marked up on arrival, a genuine value-creation lever, not narrative. It is also explicitly conditional, and the conditions are owned in full by the bear case: the lever pays only if integration delivers the EBITDA and the market keeps the multiple. De-rate the vehicle toward the tape it bought from and the arbitrage inverts, that inversion is exactly the sum-of-the-parts bear logged in the Bear section^My readEvidence. The accretion arithmetic is mine: EBITDA bought at ~10–12× marked inside a ~20× vehicle (30-Jun close basis; ~23× mid-June) transfers value on arrival, but only while the market multiple stands. Two conditions, both carried at full weight in the Bear section: integration must deliver the EBITDA, and the vehicle must keep its rating. The strongest contrary reading is that the deal tape, not the trading screen, is the honest price for this risk class, in which case the vehicle is what’s mispriced, and my own sum-of-the-parts bear (~C$5.2 blend) is that argument taken seriously.Source: my analysis over the precedent tape and the comp table; conditions cross-referenced to the Bear section..

Before the scenarios, the pond. I built the market model bottom-up rather than quoting a broker TAM^My modelEvidence. My bottom-up TAM/SAM/SOM model (v1.0, 21 Jun 2026), fleet method: annual TAM = (installed base ÷ refresh life) × content per platform + (new builds × content), for the two growth lines, MCM SAS payloads and pressure-tolerant batteries, built on a sourced platform harvest (CRS, USNI, IISS, Naval News fleet counts). SAM haircuts the Exail/Thales-captive MCM programmes; SOM applies a win-share reasoned from actual wins (Denmark, Poland, HII, Teledyne, IAI, Hanwha). Every output is an estimate; ranges wide by design; XLUUV quantities largely undisclosed. FX: USD/CAD 1.37, GBP/USD 1.343.Source: my TAM-SAM model + methodology worksheets (inputs individually cited in my evidence base); not investment advice.: the established layer, Covelya plus Kraken services, already captures ~US$315M a year and is anchored to guidance and audited accounts, not modelled^Verified anchorsEvidence. The established layer is not modelled, it is anchored: Kraken’s FY2026 guidance of C$165–175M (reiterated at Q1) plus Covelya’s audited FY2025 turnover of £142.38M (≈C$262M), a combined ~C$430M (~US$315M) of revenue already captured today.Sources: Kraken, FY2025 results + FY2026 guidance, 16 Apr 2026 · Kraken, Q1-2026 (guidance reiterated), 28 May 2026 · UK Companies House, Covelya audited accounts. The growth optionality sits in the two scaling product lines, SAS payloads and SeaPower batteries, where my base-case addressable market of ~$187M a year (range $51–553M) runs several times Kraken’s current capture, pulled by the MCM-USV recapitalisation wave and the XLUUV build-out^Reconciled · company figuresEvidence. The check that makes me trust the base case: my modelled battery SOM (~US$33M) is Kraken’s actual FY2025 battery order run-rate (~C$45M), the model ties to reality at today’s end. And the ~C$200M/yr of battery capacity Kraken says it is building sits near my TAM high (US$360M): management is sizing the opportunity at the bullish end of my own range. Capacity and order figures are the company’s.Sources: Kraken, $35M SeaPower battery sales, Jan 2026 · Kraken, investor presentation (capacity), Apr 2026.

Read it honestly: the whiskers are wide because the inputs are, a large slice of the MCM market is captive to Exail/Thales (haircut applied)^VerifiedEvidence. The captivity is primary-documented; the haircut is mine. Exail’s own rMCM record: the Belgium/Netherlands programme, twelve MCM vessels, ~one hundred drones, runs Exail’s UMISAS on the A18-M AUV and the T18-M towed sonar; Kraken content absent. Thales holds the other pole: Pathmaster selected for Singapore (first Asian export, deliveries from 2027) around Thales’s own SAMDIS NG synthetic aperture sonar.PrimaryExail, rMCM programme milestones, Belgium & Netherlands (25 Nov 2024)Thales, autonomous MCM system for the Republic of Singapore Navy (May 2025)CorroborationNaval News (IMDEX Asia 2025), Pathmaster to Singapore, first exportCaptive is not permanently closed, refits and follow-on tenders reopen; my SAM haircut treats these fleets as unwinnable in the base case., and XLUUV build quantities are mostly undisclosed. But the base case asks for no heroics: it ties to the battery orders Kraken already books today^Verified · ordersEvidence. The booked battery record: US$34M of SeaPower orders reported Feb 2025, including US$31M from an unnamed defence customer; further “major SeaPower battery orders for deep-sea UUVs” (Apr 2025); global defence orders alongside the battery-production expansion (Mar 2026); and the Dartmouth NS plant took orders before it opened (Jan 2026).Trade press (independent)Military Embedded Systems, US$34M battery orders incl. US$31M unnamed defence customer (27 Feb 2025)Unmanned Systems Technology, major SeaPower orders for deep-sea UUVs (Apr 2025)Unmanned Systems Technology, global defence orders + battery-production expansion (Mar 2026)ContextCoStar News, Dartmouth NS battery plant ‘nearing completion’, orders ahead of opening (14 Jan 2026)Order values are company-reported through trade press; customers mostly unnamed, concentration is unverifiable from outside., while the company builds capacity for the top of my range.

Against the 30-Jun close of C$6.34: bear ~C$4.60 (−27%; prime-style ~18× on C$100M EBITDA), base ~C$7.65 (+21%; ~25× on ~C$118M), bull ~C$11.25 (+77%; ~33× on ~C$130M as the XLUUV ramp and synergies land); the corners were struck against a C$6.85 reference price when the model was built^My modelEvidence. Corner arithmetic on the grid’s own formula: 33× × C$130M − C$100M net debt ÷ ~372M shares ≈ C$11.3; the bear corner (18× on C$100M) lands ≈ C$4.6 the same way. The premise carrying the bull multiple is the XLUUV ramp, so I card that premise both ways.The XLUUV premise, both waysTWZ, what the Navy’s Orca XLUUV is actually capable of (programme Q&A, Jan-2025)Strongest contrary: Defense Daily, US Navy draft memo to cancel/redirect Orca XLUUV (Oct-2025), against the May-2026 shipbuilding plan funding 16, the ramp is a live dispute, not a given.Reference-price honesty: the corners are struck against C$6.85; at the 30-Jun close of C$6.34 the same corners imply roughly −27% / +21% / +77% (my arithmetic, dated).Source: my valuation model. Estimate; not investment advice.. The grid makes the surface explicit, pure arithmetic, roughly linear between corners:

Inverting the grid is the discipline: at the 30-Jun close of C$6.34, holding a peer-appropriate ~25×, the market is paying for roughly C$98M of combined Adjusted EBITDA, the bottom of what the two companies should produce together (the mid-June C$7.12 print implied ~C$110M). The price, in other words, no longer even pays for the base case^My modelEvidence. Implied-expectations inversion on the same arithmetic; cross-checks: sell-side consensus ~C$10.80 sits between the model’s base and bull (Street span: ATB Capital Markets Sell C$6.50 ↔ Desjardins C$14 on ~35× 2027E); ownership ~72% retail / ~24% institutional, the debate is visibly unresolved.Sources: my valuation model · Desjardins via Cantech Letter · Financial Post, coverage/consensus.

The same inversion as a ladder, at the 30-Jun close^My calc · basis statedEvidence. Implied EV/EBITDA = (price × ~385M FD shares + ~C$26M net debt) ÷ EBITDA basis, at C$6.34 (30-Jun-2026 close), the same base as the gauge above. EBITDA bases: trailing combined run-rate C$88.5M (Kraken FY2025 audited C$25.0M + Covelya ~C$63.5M, a management run-rate figure, not separately audited); FY2026E combined ~C$108.5M (guidance midpoint + same Covelya figure); FY2027E base ~C$133M (my sum-of-the-parts build).Source: my calc; inputs from Kraken FY2025 results + my SOTP worksheet (evidence base). Not investment advice.:

The grid’s second use is the trade-off it makes visible. Moving one multiple turn at a fixed EBITDA is worth about C$0.28–0.36 a share anywhere on the table; moving C$10M of EBITDA at a fixed multiple is worth about C$0.48–0.89, call it two turns^My calcEvidence. Pure arithmetic on the grid’s own formula. One multiple turn at fixed EBITDA = EBITDA ÷ ~372M shares: C$105M → ~C$0.28, C$135M → ~C$0.36 a share. A C$10M EBITDA move at fixed multiple = 10 × multiple ÷ ~372M: 18× → ~C$0.48, 25× → ~C$0.67, 33× → ~C$0.89. The KATFISH reference: FY2025 guided C$120–135M, delivered C$102.2M, with management attributing the gap to programme timing.Source: my valuation model · Kraken, FY2025 financial results. Estimate; not investment advice.. The valuation is more sensitive to the earnings swing than to the rating swing, and C$10M is exactly the band one contract-timing slip can move (FY2025’s miss was deliveries sliding right), while the multiple is the variable the TSX graduation is meant to defend. The model’s risk lives in the delivery schedule before it lives in the rating.

And the inversion has two goalposts that keep it honest. On the conservative partial-year base of ~C$82M, a peer-appropriate 25× justifies only ~C$5.2 a share; on the full ~C$122M run-rate it supports ~C$7.9. The mid-June print of C$7.12 sat roughly 70% of the way up that ramp; the 30-Jun close of C$6.34 credits closer to 40% of it^My calcEvidence. Same grid basis throughout (equity = EBITDA × 25 − C$100M net debt, ÷ ~372M shares): C$82M partial-year → ~C$5.24; C$122M run-rate → ~C$7.93. Implied EBITDA at a held 25×: the C$7.12 mid-June print → EV ~C$2.75B → ~C$110M, ≈70% of the C$82→122M ramp; the C$6.34 close of 30-Jun → EV ~C$2.46B → ~C$98M, ≈40%. On the lighter fully-diluted basis used in the ladder above (~385M shares, ~C$26M net debt) each cell shifts a few percent, but the reading, the market crediting well under half the ramp at the June close, holds. Dated prices.Source: my valuation model · PNG price history (stockanalysis). Estimate; not investment advice.. Equivalently, on the run-rate the close pays only ~20×, below the ~26× the subsea pure-play cluster anchors on, so the stock is cheap only if the run-rate number is real, and demonstrably expensive on the reported partial year^My calcEvidence. EV ~C$2.46B at the 30-Jun close ÷ C$122M run-rate ≈ 20.2× (the same arithmetic at the mid-June C$7.12 print read ~22.5×) versus my mid-June subsea pure-play comps, Kongsberg ~26.2×, Teledyne ~20.2×, Exail ~17–22× on a disputed basis. Against the C$82M partial-year the identical EV is ~30×. Both readings carried: the multiple is a function of which EBITDA you believe, which is the whole argument.Source: my calc · comps from my three-agent mid-June pass (stockanalysis, valueinvesting.io, MarketScreener; 12–16-Jun-2026, single-provider caveat carried).. The price is not paying for a bubble and not offering a bargain: it is paying for successful, on-schedule integration, and the first consolidated quarter is the print that resolves which reading was right.

The Street’s own prints straddle the same surface: the house that had Kraken at Hold with a C$6.00 target in December 2025 was at Buy C$9.50 by May 2026, while ATB’s C$6.50 Sell sits just above my bear rung^Verified · one print graded softEvidence. Canaccord Genuity (Doug Taylor) held Kraken at Hold, PT C$6.00 in a 2-Dec-2025 flash on the $12M order print; by 6-May-2026 the same house showed Buy, PT C$9.50, that second print reaches me only through an auto-generated analyst-feed item, so I grade it soft. ATB Capital Markets’ C$6.50 Sell and Desjardins’ C$14 (~35× 2027E) bracket the tape in my register.PrintsGlobe & Mail, Canaccord Hold, PT C$6.00 (2-Dec-2025)Globe & Mail, Canaccord Buy, PT C$9.50 (6-May-2026; auto-generated feed item, graded soft)Dispersion this wide (C$6.00–C$14) is itself the finding, the Street is pricing the same surface I am, corner to corner.Register: my coverage-tape log (sell-side items graded, bias noted)..

The tape itself has a shape, and it is worth putting on the record while the market marks this stock down through a spring. In every complete year since the re-rating began, 2023, 2024, 2025, the second half beat the first, and not narrowly: H1 averaged +18.2%, H2 +103.6%; per month, +2.5% against +12.2%, a five-fold gap.^My calcEvidence. My decomposition of the public daily close series into month-end and half-year returns, 2023–2025 completes (methodology: month-end closes, simple averages, no compounding tricks). Caveats at full weight: three H1/H2 observations is a sample, not a law; the pattern is descriptive, not predictive; and the selection risk is real, a soft H1-2026 is exactly when this statistic looks most flattering.Sources: StockAnalysis, PNG price history (accessed Jun-2026) · my calc. The calendar shape is just as stark: October is the strongest month (+23.9% average, positive four years out of four), then September and November; March through June are flat-to-down. 2026 has followed the script with uncomfortable precision, +24.7% in January, +6.5% in February, then four straight down months, June closing 7.44 → 6.34 for roughly −15%, leaving H1-2026 at about −1%^My calcMy calc. Monthly closes off the daily series: June 2026 closed 7.44 → 6.34 ≈ −14.8%; January +24.7%, February +6.5%, March–June four straight down months; net H1-2026 ≈ −1%. As of 30 Jun 2026.Source: stockanalysis.com, PNG.V price history, my arithmetic on the published series.. None of this is a forecast; the sample is three complete years and the mechanism is inference. But the pattern and the position agree: the market is currently selling exactly the months it has always sold, into the half of the year where this stock has always lived.

The tape has a second property worth putting on the record: liquidity. Four years ago this was a C$0.34 stock that no institution could have entered or exited in size; in June 2026 it turned over roughly 1.33 million shares a session, call it C$9M a day at June prices, my arithmetic on the daily series^VerifiedEvidence. TSXV:PNG daily price & volume series, June 2026: 22 sessions, ~29.2M shares in total, average 1,327,437 shares/day; 30-Jun printed 2,574,327 shares into a C$6.34 close, +1.60%. The dollar-a-day figure is my arithmetic (average volume × ~C$7, roughly June’s mid-range price). Same registered price-history series my seasonality work is built on (StockAnalysis; underlying data S&P Global Market Intelligence; accessed 1-Jul-2026), which also carries the ~1.4–1.5M average-daily-volume reading at its 24-Jun snapshot.Source seriesStockAnalysis, TSXV:PNG price & volume history (S&P Global data)Single data vendor, delayed prices, aggregator-grade; volumes are TSXV-line only and exclude the OTC (KRKNF) tape, so total turnover is understated.Source: registered market-data series, accessed 1 Jul 2026; averages are my arithmetic., and the last session of the half printed 2.57 million shares into a +1.6% close. That is real depth for a Venture-listed name, and still nothing like the depth of the mid-cap defence cohort this business is growing into. Almost none of that tape is short, either: 4.07 million shares sold short at the mid-May count, 1.32% of the float, 1.4 days to cover^VerifiedEvidence. NASDAQ semi-monthly short-interest data via MarketBeat, as of 15 May 2026: 4,066,021 shares sold short = 1.32% of float; days-to-cover 1.4 on ~1.47M average daily volume; prior count 4.15M (−1.9%). Aggregator-grade, and it captures the US OTC line (KRKNF) only, Canadian-venue short positions are not in this count.Source: MarketBeat, KRKNF short interest (NASDAQ data, 15 May 2026). The gap between those two states is not a defect. It is where the flow argument lives.

Who is on the other side of that tape matters as much as its depth. The register is retail-heavy: at the last full snapshot institutions held about 24%, insiders about 3%, retail roughly 72%, and the top twenty-five holders together barely 27%^VerifiedEvidence. My fund-ownership sweep (Jun-2026), a three-agent pass over TradingView/FactSet ETF holdings, fintel.io 13F/NPORT records, SEC EDGAR filings and the managers’ own fund pages, each figure dated to its disclosure.Register splitFintel, KRKNF institutional ownership (13F/NPORT)Institutions ~24.3%, insiders ~3.2%, retail ~72.4%, top-25 holders 26.78%, 30-Apr-2026 snapshot (Simply Wall St, with Fintel; named, dated).ETF presenceLargest dollar position: Global X SHLD ~US$17.4M at 0.23% weight; highest weight: WisdomTree WDRN at 1.80% (~US$0.2M), TradingView/FactSet snapshot, 15-Jun-2026 (named, not linked). The one dedicated fund in the pipeline, Defiance’s 2× “Kraken Robotics” ETF (ticker KRK, Tidal Trust II), targets 200% of the daily KRKNF move per its SEC 485APOS/497 filings, a retail amplifier on the ADR line, not an institutional counterweight; the Payward crypto-exchange name collision persists only as a screening hazard.Active convictionBuyside Digest, Deep Sail Capital PartnersDeep Sail: largest single position, 14.3% of NAV (31-Mar-2026, Q1 letter); an undated screenshot puts it at 17.2%, logged as disputed, the verified figure is 14.3%. Mawer New Canada: ~C$43M / ~5.7M sh, 3.4% of a ~C$1.27B fund, its 5th-largest (31-Mar-2026, Mawer fund page, named). Fidelity small-cap complex: ~6.9M sh across four funds (NPORT, Jul–Sep-2025, dated, stale-flagged).The contrary, at full weight: Pender cut Kraken from 4.1% of its Small Cap Opportunities Fund (31-Dec-2025) to below its top ten by 29-May-2026 (Pender MRFP + fund page, named), a manager who was early and right, taking profits into strength. Snapshot dates span Sep-2025 to Jun-2026; ownership registers move faster than disclosures.Source: my three-agent ownership sweep over aggregator + filing records, Jun-2026; each figure carries its own as-of date.. Passive money is token, the largest ETF dollar position is about US$17M at a 0.23% fund weight, the highest weight anywhere in ETF-land is 1.8%, and the one dedicated “Kraken” ETF in the SEC pipeline, Defiance’s 2× daily-leveraged KRK, on the ADR line, is a retail amplifier, not an institutional counterweight. The conviction sits with active managers who did the work, a closed Mawer fund with ~C$43M^Verified · aggregatorsEvidence. The roster, cross-checked: MarketScreener (Jun 2026) lists Mawer as the largest institutional holder, 9.256%, 28.43M shares on a 307.67M pre-raise base, far above #2; Yahoo’s Major Holders snapshot (~Dec 2025) put institutions at ~26.95% across 31 names, ~27.67% of float. The ~C$43M is the closed Mawer fund’s reported position, held in my register from user-supplied manager commentary (Aug 2025), named, no link.AggregatorsMarketScreener, Kraken Robotics shareholding structure (Jun 2026)Yahoo Finance, KRKNF Major Holders (~Dec 2025)CaveatAggregator-grade: TSXV/OTC institutional ownership is poorly captured; MarketScreener’s 9.256% contests the >10% reading elsewhere in my record; both snapshots pre-date the March raise’s ~372M count., a long/short microcap fund carrying it at 14.3% of NAV as its largest position, Fidelity’s small-cap complex with ~6.9M shares, and one of them has already been selling: Pender trimmed from 4.1% of its small-cap fund to outside its top ten by late May. I keep that trim in view deliberately; it is the honest contrary to everything else in this paragraph.

The composition is the lever. The marginal buyer is still arriving: much of US retail could not buy this stock at all until late May 2026, when the OTC line went live on Robinhood, Public and Trading212^My readClaim. The access facts are verified; the flow consequence is my inference. KRKNF went live on Robinhood (commission-free), Public.com and Trading212 by late May 2026, before that, much of US retail had no practical way to buy either the OTC line or TSXV:PNG. Index and benchmark ownership is effectively nil because a TSXV listing sits outside the major index universes; graduation to the TSX, and any US uplisting behind it, which rests on a single soft source and is flagged as such in my Risks section, is what puts the name in front of mandates that cannot hold a Venture stock.Access, verifiedRobinhood, KRKNF live listingMichael Sikand (X), 29-May-2026, flagging the Robinhood availabilityThat new access converts into sustained demand is my inference, not a sourced fact, retail access also widens the seller pool, and the thin tape amplifies both directions.Source: broker listing pages (verified live, Jun-2026) + dated social flag; the flow argument is mine.; index and benchmark money owns effectively none of it because a TSXV listing is invisible to the committees that matter; and graduation to the TSX, with any US uplisting behind it, puts the name in front of mandates that cannot hold a Venture stock at any price^VerifiedEvidence. The rulebook is explicit: S&P/TSX Composite eligibility requires a minimum 0.04% index weight (10-day VWAP) and a float-adjusted turnover ratio of at least 0.50, reviewed quarterly (Mar/Jun/Sep/Dec, effective after the third Friday’s close), and the universe is TSX-listed only: a Venture listing is outside it at any size.PrimaryS&P Dow Jones Indices, S&P/TSX Canadian Indices Methodology (Mar 2026 rulebook)Strongest contraryGreenwood–Sammon via Alpha Architect, the index effect fell from +3.4%/+7.4% abnormal returns (1980s/90s) to statistically ~zero by the 2010sMandate access is a doorway, not a pump, inclusion economics have been arbitraged down.. A ~C$9M daily tape is deep enough for small- and mid-cap institutions to build a position, and thin enough that building one moves the price. That cuts both ways, and this spring it did, the same thin tape amplified the drawdown off the March high. But a register that is ~72% retail, ~24% institutional and effectively 0% passive, under a business that is about to double, is not a crowded trade. It is a queue that has not formed yet.

Nothing here says the stock must rise. It says the market has already conceded the base case, and is charging nothing for the bull.

This corpus was built to test a hypothesis, not to decorate one. After five-hundred-plus logged sources, a graded 64-driver bull register and a bear case given full weight^My registerEvidence. This verdict sits on my evidence base: 500-plus logged sources graded A–D across an 84-tab register; the bull case held as a 64-driver graded register, each driver carrying its counter; the bear case maintained in parallel at full weight from the start, not bolted on at the end.Source: my register, an internal audit trail; the load-bearing items are individually carded across the sections above. No public link, none composed., my judgement is plain: Kraken is undervalued. On every pillar that matters, this is a stronger business than the one the market paid C$10.72 for in March, record intake converting, ~C$97M of product orders YTD^Verified · my tallyEvidence. The ~C$97M is management’s YTD figure at the Q1-2026 release (28 May) and reconciles to separately announced releases: C$35M SeaPower (13 Jan) + C$24M defence order (17 Mar) + C$28M SeaPower/SAS (16 Apr) ≈ the C$87M cited at the FY2025 results, plus ~C$10M more by the Q1 release.Company primariesKraken, C$35M SeaPower battery sales, 13 Jan 2026Kraken, C$28M SeaPower battery & SAS orders, 16 Apr 2026Kraken, Q1-2026 results (~C$97M YTD cited), 28 May 2026The 17-Mar C$24M defence-order release carries no URL in my register, named here, no link composed. The annual sum is my tally of disclosed-value announcements only: a floor that excludes undisclosed awards and the ~C$50M/yr services book. Intake is announced flow, not audited backlog., the financing closed, every approval in hand, and yet the stock sits near C$6.34^Verified · as of Jun 2026Evidence. The company is measurably stronger than at its March-2026 high: FY2025 revenue C$102.2M (+12%) at 24.4% Adj-EBITDA with record order intake; Q1-2026 revenue +35% with guidance reiterated; ~C$97M product orders YTD by mid-2026; C$402.5M financing closed; every regulatory approval received 18 Jun 2026. All-time high C$10.72 (Mar 2026); C$6.34 at the 30 Jun 2026 close, +1.6% on the day, delayed data, stale by construction.Company primariesKraken, FY2025 results, 16 Apr 2026Kraken, Q1-2026 results, 28 May 2026Kraken, regulatory approval, 18 Jun 2026Kraken, C$402.5M offering close, 12 Mar 2026Market referenceStockAnalysis / TMX, PNG market data (delayed)The Dartmouth battery-plant ramp is company-stated (single source); per-hull Anduril content is my model, both graded in the sections above.. Tomorrow it becomes the West’s first listed full-stack subsea-autonomy platform, sonar, power, navigation, positioning and covert comms under one roof^My readEvidence. My characterisation. The close folds Sonardyne (acoustic positioning and covert comms), Wavefront, Voyis, EIVA, Chelsea Technologies and Forcys into Kraken’s sonar-and-power stack, positioning, comms, navigation, optical sensing and software joining SAS and SeaPower under one listed roof.Deal primariesSonardyne, Covelya Group pageOffshore Energy, Kraken buying Sonardyne & EIVA parent for C$615M, Mar 2026“First” is my read of a definition: Kongsberg is the cleanest full-stack rival but a diversified prime, and Exail is listed with navigation-and-robotics depth, the claim rests on the pure-play-plus-full-stack combination, not on being the only listed subsea name., embedded in Anduril’s Ghost Shark and Dive-XL programme^Verified · in partEvidence. The Dive lineage runs on Kraken content, sonar and pressure-tolerant batteries proven on Dive vehicles in joint sea trials (Kraken release), and Anduril’s Dive-XL was selected by DIU and the US Navy for the CAMP XL-AUV effort (Mar 2026), with Ghost Shark in production in Sydney.Programme recordKraken, Dive Technologies sea trials with Kraken sonar & batteriesBreaking Defense, DIU/Navy tap Anduril Dive-XL, Mar 2026Anduril, Ghost Shark factory opens in SydneyTiers differ inside this sentence: battery presence is verified; SAS on Ghost Shark is strongly indicated, not primary-confirmed; per-hull dollar content is my model, a bill-of-materials deep research I commissioned across Anduril and Kraken disclosures, procurement records and trade press (Jun 2026). at the exact moment allied navies commit to the undersea build-out of a generation^VerifiedEvidence. The demand signal is legislated, not hoped for: AUKUS Pillar II launched its first signature project (announced 30 May 2026); the FY2026 US budget puts ~US$5.3B to unmanned maritime autonomy (+~US$2.2B YoY) inside a ~US$901B NDAA; Germany’s Bundestag approved ~€50B of military orders in one December sitting within a ~€650B 2025–30 plan; Canada announced NATO’s 2% reached in March 2026.Government primariesAustralian DoD, AUKUS Pillar II first project, Jun 2026PMO Canada, NATO 2% announcement, 26 Mar 2026CoverageArmy Recognition, US$5.3B unmanned-maritime boost, 2025Bloomberg, Germany >€50B military orders, 17 Dec 2025Al Jazeera, US$901B defence bill passes, 17 Dec 2025Programmes are cancellable: an Oct-2025 Navy draft memo proposed cancelling or redirecting Orca XLUUV and GARC (Defense Daily), the tide that lifts can turn, and macro demand is not Kraken’s captured share..

The gap is mechanical, not fundamental^My viewEvidence. My integrated judgement of the full evidence base, every input traceable through the section cards above; the ~18.5× paid vs 28× base-case arithmetic is my model (Model & scenarios); the mechanical-mispricing read (screens, listing, trough-year optics) is inference, labelled as such.Source: my synthesis; load-bearing inputs in the Model & scenarios and Bear case sections.. A venture listing keeps a ~C$2B company off institutional screens^My readEvidence. At the 30-Jun-2026 close (C$6.34 × ~385M shares) this is a ~C$2.4B company still listed on the TSX Venture exchange, outside the index-eligibility and mandate screens through which much institutional money buys; my arithmetic, price accessed 1 Jul 2026.MechanicsTSX, graduation mechanicsS&P DJI, index methodologyFintel, KRKNF institutional registerMy read on flows: screens explain an absence of buyers, not a presence of sellers, and uplisting is optionality, not a scheduled event.; a trough-year P/E north of 650×^My calcEvidence. FY2025 net income was C$2.86M (diluted EPS C$0.01). At the C$6.34 close the simple screen is ~630× on EPS and ~850× on the ~C$2.4B market cap, “north of 650×” sits inside that basis range. The denominator is the artefact: trough-year earnings carrying acquisition and non-cash items (2024’s C$20.1M was itself flattered by a one-off C$9.7M deferred-tax benefit). Net income is the least reliable line here; EV/EBITDA is the honest lens. Price as of 30 Jun 2026.Sources: Kraken, FY2025 results, 16 Apr 2026 · StockAnalysis, PNG price history (delayed) frightens the lazy filter while the honest lens, ~18.5× FY2027E EV/EBITDA paid today against a 28× base case, for a business compounding 30%+ at ~25% margins^Company claimEvidence. Management’s standalone FY2026 guide, 30%+ revenue growth at a ~25% Adjusted-EBITDA margin, set at the FY2025 results (16 Apr 2026) and reiterated at Q1-2026 (28 May 2026), against 24.4% delivered in FY2025. Forward-looking company claim, not delivery.Sources: Kraken, FY2025 results, 16 Apr 2026 · Kraken, Q1-2026 results, 28 May 2026, now sits on the bear-case rung itself; and merger risk is being priced as if integration were novel to either side, when acquisition is precisely how both companies were built, Kraken through a decade of bought-then-scaled capabilities^VerifiedEvidence. The acquirer’s own stack was bought, then scaled: the 2016 Kraken Power purchase became SeaPower; PanGeo Subsea built the services arm; 3D at Depth (2025) added LiDAR; and the Dive Technologies relationship carried Kraken content into what became Anduril’s XL fleet. Corroborated by a three-agent M&A track-record pass I ran across Kraken’s five prior acquisitions (PRs, MD&A, audited financials, 2008–2026; no public link).RecordEE Power, Kraken acquires pressure-tolerant battery business, 2016Defense One, Anduril buys Dive Technologies, Feb 2022Scale honesty: every prior Kraken deal was a ≤C$24M tuck-in, Covelya is of a different order, which is exactly why integration leads my bear case., Covelya as a purpose-built six-brand roll-up by a proven serial integrator^VerifiedEvidence. Covelya was assembled deliberately: Sonardyne at the core, with Wavefront, Voyis, EIVA, Chelsea Technologies and Forcys grouped under one holding and presented as a coherent group, the C$615M take-out (C$480M cash + C$135M shares) buys the roll-up whole.Deal recordSonardyne, Covelya Group pageUST, Kraken–Covelya definitive agreement, 3 Mar 2026Offshore Energy, deal structure coverage, Mar 2026“Proven serial integrator” is my read of the Sonardyne-side record; transformational-M&A base rates (~70–90% failure for deals of this relative size) and undisclosed retention terms are carried at full weight in the Bear case.. The June de-rating, in other words, did the bears’ work for them: at the 30-Jun close the market pays ~20× run-rate against the ~26× the closest pure-plays command, credits barely 40% of the combined EBITDA ramp, and sits within cents of the lone Sell target on the Street^My calc · at the closeEvidence. All figures at the 30-Jun-2026 close of C$6.34, dated: implied run-rate EV/EBITDA ~20.2× against the ~26× subsea pure-play cluster; implied combined Adjusted EBITDA ~C$98M, about 40% of the C$82→122M ramp; ATB Capital’s C$6.50 is the Street’s lone Sell target; scenario corners −27% / +21% / +77% against the same close. Every figure is carded in full in the Model and Ownership sections above; my model output, not advice.Sources: my valuation model (Model & scenarios) · StockAnalysis / TMX, PNG close, 30 Jun 2026. From here the base case is +21% and the bull +77%, while the rungs below are ones the tape has already tested: the asymmetry got better, not worse, as the closing risks expired.

What would have to be true. For the bull case: the integration holds (people retained, brands coherent), the standalone book converts through H2 2026 into the combined guide^Company claimEvidence. Management’s construction: standalone guidance reiterated at Q1-2026 (revenue +35% YoY, product revenue +50%, Adjusted EBITDA positive), with combined guidance promised at closing.Company primariesKraken, Q1-2026 results, 28 May 2026Kraken, FY2025 results, 16 Apr 2026Strongest contrary in my corpus: Kraken reiterated C$120–135M revenue guidance through 2025 and delivered C$102.2M, roughly −8% organic once 3D at Depth is stripped out. Conversion is the load-bearing assumption, and it failed recently., the moat holds at its two contested edges (merchant SAS against an opened HISAS^VerifiedEvidence. Kongsberg opened HISAS to non-HUGIN platforms in February 2026, a direct attack on the merchant white space Kraken occupies. From the competitive-landscape deep research I commissioned (a full competitor map built across vendor announcements, procurement records and trade press, Jun 2026). The edge that remains is SWaP-C, open vehicle-agnostic integration, COTS pricing and the SeaPower bundle, not raw acoustics, where HISAS and Thales’s SAMDIS are peer-or-better.Source: my commissioned competitor-landscape research (methodology as stated); vendor primaries carded in Technology and Bear case.; batteries against make-versus-buy)^My readEvidence. The make-versus-buy risk is visible in the open: Anduril has been hiring battery-systems and BMS engineers in Costa Mesa with job ads that reference NAVSEA S9310, the Navy lithium-safety gauntlet any in-house pack must pass.Buyer-side signalsAnduril, Battery Systems Engineer posting (Greenhouse), 2025Anduril, Energy Systems Engineer, Maritime postingCertification barNAVSEA, Lithium Battery Safety Program guide (SBIR)It cuts both ways: S9310 is a multi-year barrier that favours the incumbent supplier, and SeaPower itself carries no public S9310 certification, a gap logged on my bear side too., and the multiple merely holds while earnings catch up. For the bear case: any one of those fails loudly, or nothing fails and the multiple compresses anyway. Both lists stay graded in the register, and the next four quarters adjudicate. The difference between this document and a pitch is that it tells you, in advance and in writing, exactly how it fails.

The verdict is my own, labelled, like every inference on this page, and it is research, not investment advice. But it is not a hedge. The work says the machine below the surface is real, the demand signal is arriving, and the price has not caught up. From tomorrow, the market gets its chance.