Week 36  |  August 2026

Monday, 6 July 2026. Inside a single trading day, three of the largest names in European and American defence announced they were buying their way into the undersea domain. Thales moved on Exail. Lockheed Martin moved on Ultra Maritime. Fincantieri moved on four companies at once. Three announcements, hours apart, each one a bet that the future of maritime autonomy belongs to whoever owns the most of it.

Rewind four days. Kraken Robotics had just closed its C$615 million purchase of Covelya, the group that owns Sonardyne. Fast-forward two days from 6 July and the pattern looked less like coincidence and more like a stampede.

Then the part the coverage mostly missed.

In the same fortnight, a different set of companies was raising money to prove the opposite point. Kraken Technology Group, a British startup with no relation to the Canadian Kraken beyond the name (more on that headache shortly), closed US$175 million at a billion-dollar valuation on 9 July. The day before it was reported it had airdropped one of its uncrewed boats out of the back of an A400M transport aircraft. And the whole time, the US Navy was running a procurement experiment called the Golden Fleet marketplace, built on the explicit premise that it should never let a single prime own the stack at all.

Two theories of how autonomy scales. Both funded, both contracted, both demonstrated, inside the same two weeks. This piece is about which one capital is actually right about.

The consolidation wave and its counter-bets, across one fortnight.

What they actually bought

Start with what these deals are not, because the trade-press framing gets it backwards.

Every acquirer described its purchase as buying capability. That reading falls apart the moment you look at what each company already made. Thales builds inertial navigation systems. So does Exail, and the two had already been picked together by the French procurement agency (DGA) to supply mine-hunting drones to the French Navy. Lockheed makes sonar. So does Ultra, and both have been selling into the same hulls for years. Kraken Robotics makes synthetic aperture sonar and subsea batteries. Sonardyne makes the acoustic positioning that tells those platforms where they are. Nobody in this wave bought a technology they couldn't have built themselves.

What was bought was structure. Production lines already certified. Supply chains already qualified. Order books already sitting inside a programme of record. Engineering teams that had already done the years of unglamorous integration work that turns a promising prototype into something a navy will actually field.

That distinction goes to the heart of what we have argued since inception. In maritime autonomy, the binding constraint has never been the technology. It has rarely even been the capital. It is industrial structure, and the certification and procurement machinery wrapped around it. These five deals are the clearest evidence yet that the primes agree. What they are buying is time.

Which sets up the only question worth asking about a wave like this. When you fold a fast-moving autonomy specialist into a defence prime, does the speed survive the absorption? Hold that thought. We come back to it, because it is where the whole bet is won or lost.

The wave, in one view

Here is what landed, and how each deal is structured.

Acquirer

Target

Value

Announced / Closed

The bet

Regulatory status

Kraken Robotics (Canada)

Covelya Group (Sonardyne, EIVA, Voyis, Chelsea)

~C$615m

Closed 2 Jul 2026

Sensor-house roll-up

Cleared (approvals obtained mid-Jun)

Thales (France)

Exail

~€3.9bn EV

Announced 6 Jul 2026

European sovereignty play

Pending: stage one to Q3 2027; tender to early 2028

Lockheed Martin (US)

Ultra Maritime (from Cobham Ultra / Advent)

US$3.45bn

Announced 6 Jul 2026

Prime absorbs a specialist

Pending: CFIUS + UK NSIA expected, not yet public

Fincantieri (Italy)

Next Geosolutions, WSense, Graal Tech, Defcomm

~€600m initial (>€1bn all-in)

Announced 6 Jul 2026

State-backed champion, built from parts

Pending: Italian golden-power a stated condition

NORBIT (Norway)

Water Linked

~US$34m

Signed 22 Jun 2026

Bolt-on

Small all-cash deal; no material review flagged

Group them by the kind of bet each represents, because they are not all the same animal.

The sensor-house roll-up. Kraken Robotics, a Canadian company listed in Toronto, spent C$615 million to buy Covelya, parent of Sonardyne, EIVA, Voyis and Chelsea Technologies. Kraken already had the sonar and the subsea power. Now it has the acoustic navigation, the survey software and the optical sensing to go with it. CEO Greg Reid has been open that the goal is a single full-stack subsea intelligence house. The deal closed 2 July, after clearing its regulatory approvals in mid-June.

The European sovereignty play. Thales agreed to buy the Gorgé family's controlling stake in Exail at an enterprise value near €3.9 billion (a 44 per cent premium) then to tender for the rest. It beat Safran, which had been in exclusive talks days earlier at a lower price and walked. Exail is widely described as Europe's leading maker of mine-countermeasure robotics and the world's number-two player in naval inertial navigation (per Investing.com). For Thales, this is about owning the sovereign French capability chain from the sensor all the way to the vehicle. Patrice Caine pitched it around European technological sovereignty, which tells you exactly which audience the deal was written for.

The prime absorbing a specialist. Lockheed Martin agreed to pay US$3.45 billion for Ultra Maritime, carved out of the private-equity-owned Cobham Ultra. Ultra makes sonobuoys and torpedo-defence systems, and it folds directly into Lockheed's Rotary and Mission Systems division. This is the purest own-the-stack move of the five. As we'll see, it is also the one that will tell us the most about whether absorption slows a specialist down.

The state-backed champion, built from parts. Fincantieri, majority-owned by the Italian state investor, spent an initial €600 million (rising past a billion once a tender offer completes) to buy into four companies at once: Next Geosolutions, WSense, Graal Tech and Defcomm. It bolted them onto its existing torpedo, sonar and submarine units to create eight "centres of excellence" under one underwater division. One number is the tell. Fincantieri now expects that underwater business to hit revenue targets it had set for 2030, and to hit them four years early. This is a landgrab.

And the bolt-on. NORBIT of Norway paid around US$34 million for Water Linked, a small Trondheim maker of underwater positioning and sensing. Tiny beside the others. Included because it rhymes: even the mid-sized sensor players are being swept up.

One footnote for completeness. Stretch the window back to 1 March and Rheinmetall's completed takeover of the German shipbuilder NVL makes it six, not five. The clustering is real either way.

The friction almost nobody wrote about

The primes can sign these deals in an afternoon. Getting them approved is another story, and the approval process is where national governments quietly reassert control over exactly the capability the prime thinks it just bought.

Take Thales and Exail. A French company buying a French company, waved through by a French government that owns roughly a quarter of the acquirer. You'd expect a formality. It isn't. The first stage is not scheduled to complete until the third quarter of 2027, with the full tender offer running into early 2028. Fifteen months, minimum, for a domestic deal with political wind behind it. The procurement agency's comfort is not in doubt, the two firms were jointly picked for the French Navy's mine-warfare programme in late 2024, but comfort and speed are different things, and a European merger review still sits somewhere in that timeline.

Then the sharp one.

When private equity bought Ultra's British parent back in 2022, the UK government did not simply nod it through. The Business Secretary issued a public-interest intervention notice, and rather than send the deal to a full competition investigation, the state accepted a set of binding undertakings. It forced the creation of ring-fenced British entities, so-called SecureCos, to hold the most sensitive capability. It put a government-appointed director on each board. And it locked down what it called "restricted information" so it could not be moved or disclosed outside the British group, or outside the country, without prior written approval. In plain terms: the government let the owner buy the company, then legally fenced off the part of the company that actually mattered and kept its own key to the fence.

Those undertakings don't simply carry over. Under the deed itself, they lapse automatically the moment Ultra changes hands, unless the UK government steps in and requires Lockheed to accept an equivalent package as a condition of closing, or opens a fresh national security review. Nobody has yet said publicly which of those happens. That is the most important thing to watch in this entire wave, and here is why.

It is the cleanest test we are going to get of the institutional-gap thesis in the wild. A prime can spend three and a half billion dollars to own an autonomy and anti-submarine specialist, and a government can still turn around and dictate how much of it the prime is allowed to run. Ownership and control are not the same thing in defence. The space between them is filled with lawyers, undertakings and appointed directors, and that space is precisely where acquired speed tends to go to die.

The others carry their own conditions. Fincantieri's Italian deals sit under Rome's golden-power screening. Kraken's own filings itemise the clearances in full, UK NSIA, CFIUS, Danish FDI screening, and ITAR/DDTC conditions. None of these is a rubber stamp. Every one is a point at which doctrine gets a vote.

The other bet: open it

So much for owning everything. The companies wagering the other way are not fringe holdouts. They are funded at the same scale, and they have something the primes do not: a customer writing the rules in their favour.

Deal with the naming problem first, then set it aside. Kraken Technology Group (KTG) is British, founded by a former offshore powerboat racer, and has nothing to do with Kraken Robotics of Canada beyond a shared mythological animal. KTG’s model is the mirror image of Fincantieri's. It owns the design of its K3 Scout uncrewed boats and the autonomy integration, and it rents the manufacturing. Series production runs through a joint venture with Rheinmetall in Hamburg. A separate partnership with Anduril handles American production. Another, with Davie, covers Canada. The autonomy software comes from Auterion, an open operating system, not something KTG built in-house and guards. When the company raised its US$175 million in July, at a valuation that made it Europe's newest defence unicorn, the pitch was reach and speed without the capital anchor of owning steel.

RAD Propulsion, another British firm, makes the same philosophy explicit at the component level. Its Axon system, launched in May, is an open architecture for uncrewed vessels whose entire selling point is that it does not lock the customer into anyone's autonomy or mission software. The propulsion layer and the brains are deliberately kept apart, so each can evolve on its own clock. Co-founder Rich Daltry has said flatly that the industry is moving toward open, flexible architectures where the software and the mission applications develop independently of the hull. That is a shot at the own-the-stack thesis, fired from inside the same market.

Here is what gives the open camp real weight, though. It isn't only a commercial preference. It is increasingly the law. The US Department of Defense mandates a modular open-systems approach, and the US Navy runs its own interface standard for uncrewed maritime autonomy, built specifically so payloads and software from different vendors plug into the same hull. There is an allied logic layered on top: interoperability across AUKUS partners is far easier when everyone's autonomous systems speak a common, open standard, which is exactly the direction Pillar II keeps pushing. When the customer legislates against vendor lock-in, the vertically integrated prime is selling the one thing the buyer has decided it doesn't want to buy.

And then there is Saronic, which refuses to sit on either side. It is building its own shipyard, a US$3.2 billion site in Brownsville, Texas, which sounds like the most vertical move imaginable. Yet it does it as a venture-backed disruptor valued at US$9.25 billion, not as a legacy prime, chasing the same contracts as the partnership-model players. Kleiner Perkins, which led its latest round, put the logic plainly: maritime dominance needs production capacity as much as it needs technology, and the two rarely turn up in the same company. Saronic's bet is that it can be the exception. It scrambles the neat binary this whole piece is built on, and it earns its place in the frame precisely because it does.

Own it vs open it: where the July 2026 money placed its bets. Positions illustrative, not a precise ranking.

Does the speed survive?

Back to the question we parked earlier. Fold a nimble autonomy specialist into a prime, and does its speed survive?

Honest answer: it is too early to know for these five, because the ink is barely dry. But you can read the intent from the integration structures, and they differ in ways that matter.

Lockheed is folding Ultra straight into an existing division. Full absorption. That is the structure most exposed to the risk that a fast specialist inherits the procurement rhythm of a company built around the F-35. Fincantieri went the other way, keeping founders attached through earn-outs that run as far as 2031, which is a deliberate way of saying stay, keep building, do not slow to our pace just yet. Kraken Robotics brought Covelya's leaders onto its own executive team rather than dissolving them into it. Same wave, three different theories of how not to smother the thing you just paid for.

The sceptic's case has a natural spokesman in Anduril, whose whole origin story is that legacy primes deliver slowly and expensively because the contracts reward exactly that. There is truth in that. There is also an obvious interest behind it, given Anduril is now a prime itself and competes with nearly everyone in this piece. Take the critique seriously, and weigh the source.

The other side of the argument deserves equal air, because it might be the right one. Vertical integration could genuinely accelerate autonomy rather than smother it. The real bottleneck, the thing we keep circling back to, is access to funded programmes of record and the certification pipelines that gate them. A startup can burn years trying to get through that door. A prime already owns the door. Thales does not need to talk its way into the French mine-warfare programme because it runs it. Owning Exail closes the loop from sensor to vehicle to navy inside a single company that is already inside the tent. If speed dies in the procurement queue, then handing a specialist a fast pass to the front of that queue might be the most speed-preserving thing you can do for it.

We don't think this one is settled, and we’re suspicious of anyone who says it is. The track record of primes absorbing fast-moving acquisitions is genuinely mixed, and we’re not going to pretend the history points cleanly one way. Both stories are coherent. Fielded hulls, not press releases, will separate them.

The buyer built a bazaar

Now let’s widen the lens to the customer, because the most interesting collision here is not between two groups of sellers. It is between the sellers consolidating and the way the biggest buyer has decided to shop.

The US Navy relaunched its medium uncrewed surface vessel effort in 2026 as a "marketplace," under the Golden Fleet banner. The design principle is the opposite of own-the-stack: qualify as many vendors as possible, skip the years of prototyping, get capability on the water fast, and favour modular payloads that move between hulls. It is the open-architecture theory expressed as procurement doctrine, and it is running at the exact moment the supply side is consolidating into a handful of full-stack owners.

The friction is already showing. In mid-July, both Saildrone and Blue Water Autonomy went to the Court of Federal Claims over their exclusion from that marketplace, each arguing it had been shut out prematurely. Build an open bazaar, start turning vendors away at the door, and the vendors sue. That is the institutional gap generating litigation in real time.

The two stories have also started to physically touch. Ultra Maritime, the target of Lockheed's pending US$3.45 billion acquisition, brought a counter-drone sonar called Sea Spear to the Navy's Lanternfish exercise in late July, independently, roughly three and a half weeks after the deal was announced and still well before any close. It was detecting and classifying uncrewed undersea vehicles, working alongside a seabed sensing system from, of all companies, Anduril. So the target of a deal that just bought its way into the stack, on paper, is field-testing next to the open-model disruptor its future owner competes with everywhere else. The categories are blurring at the waterline, even before the ink is dry.

What capital is really buying

What is capital rewarding, then, if both models are raising billions?

Both, is the uncomfortable answer, for different buyers. The advisory houses tracking defence M&A this year describe investors paying premiums for portfolio clarity: carve-outs, pure plays, vertical integration that de-risks a supply chain. That is the money behind Thales and Lockheed. At the same time, venture funding for defence technology hit a record last year, well north of the previous high, and a large share of it is flowing to exactly the asset-light and build-your-own-yard plays that reject the roll-up logic. Kraken Technology's unicorn round and Saronic's nine-figure raises come out of that same pool.

The market, in other words, is long both theories at once. What it is actually pricing, underneath the strategy labels, is a single scarce thing: the ability to put fielded capability on the water at the pace the doctrine now demands. The vertical roll-up and the asset-light partnership are two competing claims on that same quality. Capital is funding the race, not calling the winner.

The OTI Take

Five deals in a month, six if you count Rheinmetall. Read together, they are a bet on industrial structure. Every acquirer already had the engineering. What they lacked, and what they paid billions to buy, was the certified production and the programme access that turns autonomy from a demo into a delivered capability. On our thesis, that is the primes conceding the point we have made since inception: the constraint was never the technology.

But the counter-bet is just as well funded, and it holds the one asset the primes cannot acquire: a customer legislating in its favour. Open architecture has stopped being a preference. In the US Navy it is policy. That is a structural headwind against every company that just spent billions to own a closed, integrated stack.

The deciding test is sitting in a British government file. Whether the UK government forces Lockheed to accept fresh undertakings equivalent to Ultra's 2022 SecureCo package, or lets the deal close without them, is the cleanest natural experiment we're going to get on whether a prime can absorb a specialist without a government fencing off the good part. Watch the UK investment security unit. That, more than any deal value or press release, will tell you whether owning the stack delivers control, or just the invoice.

The primes are betting they can buy their way to undersea autonomy at scale. Their rivals are betting the buyers just made themselves slower and dearer at the precise moment the customer started paying for the opposite. Both cannot be right. The water will settle it, and probably sooner than the eighteen-month contract cycle would suggest.

Next week

The MASS Code has been in force since 1 July: though it's voluntary for now, a binding version isn't due until the early 2030s. A master must remain responsible for the vessel at all times. Doesn't need to be aboard.

Fine in principle. Then ask which regulator licenses that master, whose law governs when the link drops, and what happens when the master sits in Norway, the operations centre sits in Singapore, and the hull is flying a Maltese flag through UK waters.

This week was about who owns the stack. Next week is about who answers for it.

Since you have been, thanks for reading.

Cheers,

Mick

Ocean Tech Intelligence provides informational analysis only. Nothing in this publication constitutes financial, investment, legal, or strategic advice. Readers act on this content at their own risk. For full details see our Disclaimer.