DEEP DIVE: Own the Factory or Own the Sonar

Week 40  |  September 2026

Six weeks. One buyer. Four companies moving in opposite directions.

Between the middle of July and the end of August, Saronic committed to a third shipyard, signed a Korean production-technology partner, and opened a commissioning site on the Gulf coast. In the same window, Anduril quietly walked out of a shipyard it had spent tens of millions of dollars revamping. Leidos left its industrial base exactly where it was. All four sell autonomous surface vessels to the United States Navy (Navy). The Navy has not said which way of building them it prefers.

That silence is the story.

Saronic moved three times on owned capacity while Anduril shed a yard and Leidos held still. One Navy, one window, opposite bets.

As we’ve been saying for a while now, the technology question is settled, or close enough to it. Back in July, we analysed the three-Saronic-Corsair-strike on a naval facility at Bandar Abbas, the first time American forces have used sea drones in combat. The platforms work. What isn't settled is how you build them at the scale the Navy keeps saying it wants: not tens of hulls, but hundreds. And there is no version of the United States shipbuilding base, as it exists today, that was sized to deliver that on either of the two models now competing to try.

Call them what they are. One theory says you own the factory. You build the yards, you pour the concrete, you control the line end to end, and the capacity you own becomes the thing nobody can take from you. The other says you own almost nothing that floats. You own the smart content, or the integration, or the design, and you let other people carry the steel. Both theories are live, and both are funded. They are also both being bought right now by the same customer, who is behaving as though it does not need to choose.

It will have to. The question we ask in this deep dive is what happens when it does, and who is carrying the risk while it waits.

Back in Week 36 we mapped the whole consolidation wave, the run of deals that reshaped the sector across a single summer. This is the narrower cut: two companies, two incompatible theories of how you actually reach volume, followed all the way down. Saronic at one pole. Kraken Robotics at the other. And in between, the field that most of the industry has quietly chosen.

If you build autonomous vessels, invest in the people who do, or buy from them, the next ten minutes are worth the read.

Let's start with why Saronic can even attempt this, because the answer is money.

In March, the company closed a USD1.75 billion dollar Series D at a USD9.25 billion dollar valuation, led by Kleiner Perkins. That is not a number you raise to build boats one at a time. It is the balance sheet that makes owning capacity plausible, and Saronic has spent the months since turning it into physical plant at a pace that is genuinely hard to match.

Their stack runs three deep now. Austin is the headquarters and the small-craft line, a total presence of more than 520,000 square feet. Franklin, Louisiana came in April 2025, a former Gulf Craft yard now getting a USD300 million dollar expansion with three slips and a large-vessel line. And then, on 16 July 2026, Port Alpha: a greenfield shipyard at the Port of Brownsville, USD3.248 billion dollars by the Texas Governor's figure, 835 acres with room to grow past four thousand, built to turn out vessels up to 850 feet, operational in 2028. Governor Abbott put the fully-built headcount at around 10,000 employees and the annual Texas payroll at USD750 million dollars. The state kicked in an USD80 million dollar Enterprise Fund grant to land it.

That is the vertical-integration thesis rendered in concrete and acreage. Own Austin, own Franklin, own Port Alpha, and you own the line from a six-foot scout craft to something the length of one and a half football fields.

The combat record is what earns Saronic the right to be taken seriously here, not just as a shipyard developer but as a supplier the Navy is actually leaning on. The Bandar Abbas strike on 12 July put three military-variant Corsairs against a submarine and ship-maintenance facility. CENTCOM’s language was flat and deliberate: the first time American forces have employed sea drones in combat operations. The Corsair itself is a 24-foot craft, about a thousand pounds of payload, a thousand nautical miles of range, top speed better than 35 knots. It sits on a USD392 million dollar production contract announced at the Reagan forum last December, awarded through a Defense Innovation Unit pathway rather than a traditional program of record, and then-Secretary of the Navy John Phelan made a point of the timeline: prototype to production in under a year.

Weeks before the strike, a Navy-operated Corsair helped pull two US Army aircrew out of the water near the Strait of Hormuz. The platform is doing real work in a real theatre. Nobody is arguing the tech isn't ready.

So the tempo matters. Three capacity moves in six weeks, on top of a combat deployment and a marketplace slot. On 23 July 2026, a week after Port Alpha, Saronic added Samsung Heavy Industries as a production-technology partner, robotics-based process automation and workforce know-how, at the launch of a Korea-US shipbuilding centre. In the second week of August it opened a test-and-commissioning site at the Port of Gulfport, Mississippi, for the big Marauder MUSV. Every one of those moves pulls in the same direction: build more, own more, run the line faster.

On capacity, there’s a caution the company's own numbers earn. Saronic says Austin will eventually turn out up to 2,000 Corsairs a year, and Franklin up to 20 Marauders a year. Those are company figures, not independently verified, and worth holding at arm's length until a hull count backs them. The dollar amounts, the acreage, the jobs, those check out against primary reporting. The throughput claims are the part still waiting on evidence.

The logic of the bet is clean nonetheless. Control the whole line and the factory becomes the moat. Nobody can raise your prices, throttle your slots, or walk away mid-program, because you own every step. If you believe volume is the war you're going to win, you buy the ground you'll fight it on.

There's a cost to that belief, and we'll come back to it, because it's the sharpest part of the whole argument.

Here's the thing about Saronic's approach. Almost nobody else in the sector is copying it.

Saronic and Kraken Robotics sit at opposite poles. Most of the field owns the integration or the design and distributes the build.

We flagged the pattern back in our Week 22 Brief on 4 May: across the US programs, the recurring shape is a platform specialist paired with an autonomy integrator, or a prime distributing hulls across yards it doesn't own. Not one company building everything itself. A network, deliberately spread.

The clearest recent marker is HII. On 6 July 2026, they added Halimar Shipyard in Morgan City to its ROMULUS production network, to build complete 190-foot vessels alongside Breaux Brothers, where five are already under construction. Andy Green, who runs HII's Mission Technologies arm, framed it as building the industrial capacity to deliver autonomous maritime capability at scale. HII is also standing up a new assembly facility at Breaux Brothers and has brought in Path Robotics for AI-driven welding and GrayMatter Robotics for surface prep and finishing. That is a prime choosing to distribute rather than centralise, spreading the build across partner yards instead of concentrating it in one owned megasite.

The same shape shows up everywhere you look. Hanwha's US defence arm paired with Magnet Defense in the northern spring to build a 38-metre MUSV and, in their words, AI-driven robotic shipyards, Hanwha's manufacturing married to Magnet's autonomy. Hanwha also paired up with HavocAI on a 200-foot autonomous vessel, and Hanwha Ocean brought Leidos in on naval ship design through Gibbs and Cox. Edison Chouest's Galliano Marine Services took on domestic MUSV production with Anduril, working from a hull design that started at HD Hyundai. Every one of these is the same architecture: specialist plus integrator, capacity assembled from parts rather than owned outright.

Which brings us to the six-week contrast, and it needs stating carefully.

In the same window that Saronic moved three times on owned capacity, Leidos made no change to its industrial base. No new yard, no serial-production partner, no capacity expansion, no maritime acquisition. That is not a company falling behind. It is a company whose model doesn't require concrete on a six-week cadence, because Leidos owns no shipyard by design. It builds through commercial and partner yards and runs a distributed strategy, most visibly with Sea Archer in Australia, which Leidos wants built across a dozen or more boatyards. An asset-light integrator that isn't pouring foundations is behaving exactly as the model says it should. The contrast with Saronic isn't one company working and another idle. It's two theories doing precisely what their internal logic predicts. One acquires ground. The other stays deliberately light.

That is the real division in the sector. Not a spectrum of effort. A genuine disagreement about what you need to own to win.

At the far pole from Saronic sits a company built on the opposite premises: you needn’t own hulls at all.

Kraken Robotics closed its acquisition of Covelya Group on 2 July 2026. What matters here isn't the transaction, it's what it turned Kraken into, and the answer sharpens the contrast with Saronic to a point.

Saronic's bet is that the factory is the moat. Kraken's is that the factory is a distraction, and the real defensible ground is the smart content inside everyone else's hull. Sonar. Batteries. Acoustic positioning. The systems that make an autonomous platform actually autonomous, actually useful, actually able to see and navigate and endure. Own those, embed them everywhere, and you don't need a shipyard. You need other people's shipyards, as many as possible.

Before Covelya, Kraken was essentially a two-product company: synthetic aperture sonar and pressure-tolerant SeaPower batteries, both high-margin, both embedded in platforms it doesn't build - Ghost Shark, REMUS, and allied navy programs. Covelya widens that content dramatically, acoustic positioning and navigation through Sonardyne, survey software through EIVA, optical sensing through Voyis, more sensing through Chelsea Technologies. Greg Reid, Kraken's chief executive, described the intent as making the company a global provider of dual-use subsea intelligence. In plainer terms: the full sensing and navigation payload, sold into any hull that will take it.

The economics are the mirror image of Saronic's. Kraken manufactures its sonar in Newfoundland and its batteries in Nova Scotia, with a Halifax-area expansion underway, and turns that into product sales at better than 60 percent gross margin. No slipways. No 3 billion dollar greenfield. No 10,000-person payroll to carry through a lean procurement year. Combined 2025 revenue ran around CAD365 million dollars at a 24 percent adjusted EBITDA margin, and the post-close guidance points to a business that scales by shipping more high-value boxes, not by pouring more concrete.

Its Navy relevance is real, if one step removed. Kraken content rides inside platforms the US and its allies are buying: batteries and sonar in Anduril's Ghost Shark and Dive family, sonar in the REMUS line. Estimates of exactly how much Kraken content sits in each Ghost Shark come from analysts rather than the companies, so treat the per-hull figures as indicative. The direction is what counts. As those platforms scale, Kraken's content scales with them, and Kraken never has to build a single hull to benefit.

One point of housekeeping, because the names are a trap. Kraken Robotics, the Canadian company we're discussing, is not Kraken Technology Group, the British USV builder that partners with Anduril on surface vessels. Same word, different companies, different bets entirely. When this deep dive says Kraken, it means the Canadian subsystem house.

So, hold the two poles side by side. Saronic owns the line and bets volume rewards the owner. Kraken owns the intelligence and bets volume rewards whoever's inside every hull regardless of who built it. Between them sit HII, Hanwha, Leidos, Chouest, each owning some slice, the integration or the design or the payload, and distributing the rest. Three flavours of the same underlying wager: that in a world where the Navy wants hundreds of hulls fast, flexibility beats ownership.

Saronic is betting the opposite. Which is why the next part matters so much.

Just over two weeks ago, Anduril confirmed it had walked out of the old Foss Shipyard on Seattle's Lake Washington Ship Canal. The company had spent, by its own account, tens of millions of dollars revamping the site. The explanation was blunt: the space supported a Navy program pursuit, the Navy cancelled that program, and the footprint was no longer needed.

The cancelled program was MASC, the Navy surface-drone prototyping effort the Foss yard existed to serve, since scrapped and folded into the Golden Fleet marketplace. When it died, the yard's whole reason for being died with it, and Anduril left.

Here's why that matters for everything above. Anduril's surface-vessel model is largely asset-light, built around partners like HD Hyundai and, yes, the British Kraken Technology Group. Foss was only ever low-rate production for integration and testing, a handful of employees in a small footprint. So when the program vanished, absorbing the loss was cheap. A few staff, a lease, some sunk fit-out. Gone, and the company barely broke stride, its multibillion-dollar valuation carrying on untroubled.

Now run the counterfactual. What happens to a company holding USD3.2 billion of Port Alpha concrete if a program it's counting on gets cancelled the same way MASC did?

That is the cost buried inside the vertical bet. Owned capacity is a moat and a liability at the same time. The thing that can't be taken from you is also the thing you can't put down. Anduril demonstrated the asset-light escape hatch by using it, cleanly, in public, not three weeks ago. Saronic has deliberately built a model with no such hatch, on the wager that the Navy's demand is durable enough that it will never need one.

To be fair to Anduril, and precise: this is not a company that avoids owning anything. It owns Quonset Point in Rhode Island, where it builds Dive-LD and Dive-XL undersea vehicles, and it built a purpose-built Ghost Shark facility in Sydney to serve an AUD1.7 billion Australian production contract. It owns plenty. The point is sharper for being narrow: even a company that owns significant production treated a surface-vessel yard as disposable the moment the program under it disappeared. That is the asset-light logic working as designed, not a company stumbling.

And it lands directly on the question the Navy won't answer.

The Navy is buying from both theories at once, through the Golden Fleet MUSV marketplace we covered in Week 32. Seven vendors were selected in late May for at-sea testing: Sea Machines, Leidos, Saronic, Galliano Marine Services, PacMar, Birdon and HII. Those seven span the entire spectrum. Saronic, owning its yards. HII, distributing across partners. Galliano running Anduril's hulls. The vertically integrated and the deliberately light, side by side in the same marketplace, and the Navy's stated intent is to qualify as many as it can.

It has not said which industrial structure it prefers. That is not an oversight. It is the whole institutional gap this publication keeps returning to: the capability exists, the platforms work, and the thing that isn't built is the acquisition architecture that would tell suppliers which way to bet.

Underneath sits a base that was sized for neither theory. By the White House's own February action plan, there are only eight US shipyards that can build vessels longer than 400 feet. The Navy operates 291 battle-force ships against a legal requirement of 355, and the shipbuilding budget has doubled over two decades without producing a single additional hull relative to 2003. The policy response, the SHIPS Act, the maritime action plan, leans on private capital and distributed shipbuilding, which is to say it leans on exactly the asset-light network model, while Saronic tries to buy its way clean around the constraint by building new capacity outright.

Both models are unproven at the volume Golden Fleet implies. Neither has delivered hundreds of hulls, because nobody has been asked to yet. The Navy is running a live experiment with real suppliers' balance sheets as the test apparatus, and it has not defined what winning looks like. When a program gets cancelled, as MASC did, the cost of that indecision lands somewhere. Anduril showed it landing softly on an asset-light footprint. The open question is whether it ever lands hard on an owned one, and who's standing there when it does.

Right now, today, Saronic looks like the company executing best. Three capacity moves in six weeks, a combat deployment, a marketplace slot, a balance sheet most competitors can't approach. Set against Leidos making no industrial-base move in the same window, the tempo contrast is real and it favours Saronic.

But tempo is not proof. Moving fastest on owned capacity tells you Saronic is committed, not that vertical integration is the model that reaches hundreds of hulls at a lower cost than the network. That question is genuinely open, and the honest position is that neither theory has earned the win yet. Saronic is betting durable demand justifies owned plant. Kraken is betting content ownership beats hull ownership in any volume scenario. The distributed primes are betting flexibility beats both. All three are rational. Only one buyer decides who’s right, and it hasn't.

What would settle it is a handful of markers over the next two quarters. The MUSV production awards, expected around the end of September, are the first hard read on whether the Navy rewards owned capacity or distributed networks. Port Alpha breaking ground by year-end tests whether Saronic's build timeline holds. Kraken's first combined quarter with Covelya, due in late November, tests whether the acquisition-led model delivers the margins it promised. And any move by Leidos, or another network player, into serial production would confirm the distributed model as the sector's default rather than its hedge.

Until those land, the situation is what it is. The Navy wants hundreds of hulls, has funded two incompatible ways of building them, and has declined to say which it's standing behind. Two theories of how this scales, both plausible, both exposed. The suppliers are carrying the risk of that indecision on their balance sheets, and the buyer is letting the market run the experiment.

That's a comfortable position for exactly one party in this arrangement. It isn't the one building the ships.

The yards are Korean, and the question is whose navy. Hanwha owns a Philadelphia shipyard, the US Navy has taken an RFI to South Korean builders, and Canada is weighing Korean submarines. Three separate trade stories, one condition underneath them: allied naval shipbuilding capacity now sits substantially outside the alliance's own industrial base. We look at what a foreign-owned domestic yard actually changes, and whether the acquisition rules were built to buy capability from where it now lives.

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.