Ulysses sells underwater drones at $50,000 against the Navy's $5 million

1. Seagrass, the unsung hero — ▶ 8:29

   Seagrass covers 2% of the seafloor but holds 20% of the ocean's carbon, and about 20% of commercially important fish stocks use it as habitat. Coral gets the funding; seagrass is "this unsung hero of the ocean." Restoration runs 100 to 1,000 times slower than loss. On current projections many critical fish stocks could be wiped out around the 2040s, with billions of people relying on them for protein or income. Overfishing is one driver, habitat loss the other.

2. Mothership plus modular sub — ▶ 9:40

   Leviathan, an autonomous surface vessel with a 2,000 nautical mile range, does the transit because there is less drag on the surface. It then drops Mako, the underwater vehicle. Mako is built from Lego-like modules: extra battery packs add 150 nautical miles, a different rear thruster adds speed, stereo cameras give 3D imaging, a sonar front module gives bathymetric maps. Earlier underwater vehicles mainly gathered data; Mako is designed to manipulate. That architecture let Ulysses say yes when the Navy and offshore energy companies came calling.

3. The frontier playbook — ▶ 16:53

   Frontiers develop in stages. First come expeditions justified by a single high-value prize: furs in the American West, spy satellites in orbit. Then access gets cheap, through the railroad or reusable launch, and settlement becomes persistent. O'Brien wants the ocean to make that jump. We have mapped more of Mars than the seabed, over 75% of the ocean has no eyes on it, and fewer than 100 people work on it daily. Permitting is the biggest barrier: "green tape" has stalled a potentially record-size seagrass project in Australia.

4. Verticalizing the stack — ▶ 22:21

   The first vehicle was about 80% off the shelf; Ulysses is now on generation five. Software is already written from scratch, ignoring existing actuation frameworks. Over the next year, sonar and inertial navigation, both bought today, move in-house, with an acoustics team being hired now. An early inertial navigation prototype costs a fraction of the long-lead-time unit it replaces and sits on a supply chain that can scale. His investor notes this approach would once have got a founder "shown to the parking lot."

5. Building for millions — ▶ 24:49

   About 20 million aerial drones are built each year against 1,000 to 2,000 underwater ones. O'Brien says realizing the vision means building millions, so Ulysses builds on EV, drone and consumer-electronics supply chains, not maritime ones. One supplier makes only 100 to 200 of a key component a year, while Ulysses hopes to build a few hundred vehicles next year, maybe 1,000. A factory opening next year could be among the largest underwater-drone plants by units. The company is saying no to contracts, and one of its four cofounders works only on manufacturing.

6. Corrosion, pressure, comms — ▶ 27:55

   Corrosion is manageable with coatings and anodizing. Pressure is harder: Mako has been crush-tested to 2,200 meters and typically operates at 1,500, where most economic activity happens today. Full ocean depth, 6,000 meters, is far harder. The biggest problem is comms: acoustic links manage about 10 kilobytes per second over 5 kilometers, so autonomy runs onboard, on an NVIDIA 5090 and Intel CPU, around 100 times the compute of other autonomous underwater vehicles. Behavior is still basic: draw a polygon, get a lawnmower track, go around rocks.

7. First contracts: Navy, wind — ▶ 31:16

   Year two brought the first US Navy contract, in mine countermeasures, and the first offshore wind contract. In California, floating wind power cables attract abandoned fishing lines that catch dolphins and whales, one of the biggest objections to the technology. Ulysses is building a system to map the cables daily, check their integrity, detect snagged lines and cut them free. Seagrass work continues in Virginia, Southern Florida, Sausalito and the Great Barrier Reef. Per O'Brien, new contracts need no new R&D, only more field operators.

8. Heart-led hardware founders — ▶ 34:39

   O'Brien splits founders into head types, like Jeff Bezos picking books because the internet was growing 30% a year, and heart types, like Steve Jobs. He sees this hard-tech wave as heart-driven: "make something the world needs," not just something people want. His hedged explanations: COVID hit in college and pushed founders toward the physical world, a return to organized religion, a reaction to the woke era. He lives with three Christian founders, and calls his sample small enough to need a Student's t-test.

9. Replacing the crewed ship — ▶ 38:20

   Ocean work today means a chartered ship at a few hundred thousand dollars a day, after a backlog to get one. Eight-hour shifts mean three crews per 24 hours, plus managers, captain, second mate, cook and doctor. Autonomy removes that labor, works in worse seas and goes faster, since nobody gets seasick. Mako can sell to navies for as little as $50,000, against roughly $5 million for the most common AUV the Navy buys today. With advanced sensing and navigation the gap narrows to one order of magnitude. Leviathan costs $500,000 to $1 million.

10. Slow buyers, sticky contracts — ▶ 39:47

    Energy buyers demand health and safety approvals and a master services agreement; governments demand a contract vehicle and clearances. O'Brien says venture-backed companies can fund R&D through the wait, and winners are sticky because nobody wants two years to swap in a replacement. His first startup, a scooter-sharing company in the UK and Ireland, taught him to sell to government. His thesis: the smartest people in 1960 went into government and now go to San Francisco, so "there is a future where Ulysses is the largest navy in the world."

11. Cable chokepoints — ▶ 46:01

    There are about 200 undersea cables, usually taken out by accidental anchor dragging, but a targeted demolition of many of them is not hard. Nearly all US-to-Europe cables cross off west Ireland, with similar chokepoints in Asia and the Middle East. Persistent 24/7 monitoring is too expensive manually. Ulysses has work in this area but will not discuss it. O'Brien says defense peers focus on making things go boom; the hard problem is a general-purpose underwater platform, which he says turns a $3 trillion ocean economy into a $10 trillion-plus one.

The ocean is expensive to work in because every task is done the same way, with a crewed ship. Ulysses exists to take the ship out of the loop. A chartered vessel costs a few hundred thousand dollars a day, and getting one means joining a backlog. Aboard are operators on eight-hour shifts, so a 24-hour operation needs three of them, plus managers, a captain, a second mate, a cook and a doctor. Whether the job is finding a mine, repairing a pipeline or planting seagrass, the shape is identical. Remove the crew and the labor cost goes with it. The vehicle can also work in worse seas and move faster, because nobody on board gets seasick.

The price gap is the pitch. Ulysses can sell its Mako underwater vehicle to a navy for as little as $50,000. The most common autonomous underwater vehicle the US Navy buys today costs around $5 million. That is two orders of magnitude on the base vehicle. Add advanced sensing and navigation and the gap narrows to one. The Leviathan surface vehicle, which carries Mako up to 2,000 nautical miles, charges it wirelessly and links it to Starlink, costs between $500,000 and $1 million. Offshore energy customers do not buy vehicles at all. They buy services, which Ulysses runs itself.

A price like that cannot come from maritime suppliers. The world builds 1,000 to 2,000 underwater drones a year and 20 million aerial ones. Will O'Brien wants Ulysses to ride the supply chains of electric vehicles, drones and consumer electronics, where volumes run to millions. So the company is pulling its stack in-house. The first vehicle was about 80% off the shelf. The current one is generation five. Software has always been written from scratch. Sonar and inertial navigation, both bought today, move in-house next year, and an early navigation prototype costs a fraction of the long-lead-time unit it replaces. One component supplier makes 100 to 200 units a year. Ulysses hopes to build a few hundred vehicles next year, maybe 1,000. Demand is not the bottleneck. The company is turning contracts away. One of its four cofounders does nothing but manufacturing, and a factory opening next year could be among the largest underwater-drone plants in the world by units. Pitch that to a venture capitalist a decade ago, his investor notes, and you would have been "shown to the parking lot."

The design follows the same logic. Mako is built from modules. Extra battery packs add 150 nautical miles, a different rear thruster adds speed, stereo cameras give 3D imaging, and a sonar front module gives bathymetry. One form factor serves the Navy, offshore energy and seagrass restoration. That is also how O'Brien separates Ulysses from other defense startups. Many of them are working out how to make things go boom. He calls that the uninteresting problem. The hard one is a general-purpose underwater robotic platform, and that, he says, is what turns a $3 trillion ocean economy into a $10 trillion-plus one.

The company's origin explains why the platform had to be general. It began when a cofounder met a marine biologist whose job was photographing, planting seeds and collecting water samples by hand. Seagrass covers 2% of the seafloor but holds 20% of the ocean's carbon. Restoration runs 100 to 1,000 times slower than loss, and many critical fish stocks could collapse around the 2040s. Mapping, collecting and planting turned out to be a template for a dozen other underwater jobs.

Behind it sits a theory of how frontiers open. First come expeditions, each justified by one high-value prize: furs in the American West, spy satellites in orbit. Then access gets cheap, through the railroad or reusable launch. Then settlement becomes persistent. O'Brien wants the ocean to make that second jump. More of Mars has been mapped than the seabed, and over 75% of the ocean has no eyes on it. Undersea cables show what persistence would mean. There are about 200, usually severed by accidental anchor dragging, though a targeted attack on many of them would not be hard. Nearly all the cables from the US to Europe cross off the west coast of Ireland. Patrolling such chokepoints around the clock is too expensive with crewed ships. Ulysses has work in this area. O'Brien will not discuss it.

First revenue is concrete. The Navy contract covers mine countermeasures. The offshore wind contract targets a specific obstacle in California. Power cables on floating turbines attract abandoned fishing lines, the lines catch dolphins and whales, and that is one of the biggest objections to the technology. Ulysses will map the cables daily, check their condition, spot snagged lines and cut them free. Seagrass work continues, and new contracts need no new R&D, only more operators in the field. Selling is slow. Energy buyers want health and safety approvals and a master services agreement. Government wants a contract vehicle and clearances. O'Brien answers that venture money is built for exactly this. It can fund R&D through the wait, and a winner is sticky because nobody wants to spend two years bringing in a replacement.

The obvious objection is that millions of vehicles need a market that does not yet exist. Today's buyers are a few navies and one energy developer. The autonomy is, in O'Brien's words, "relatively rudimentary": draw a polygon, get a lawnmower track, go around a rock. Acoustic comms run at about 10 kilobytes per second over 5 kilometers, so everything runs on the vehicle, which carries an NVIDIA 5090 and roughly 100 times the compute of other autonomous underwater vehicles. Mako operates to 1,500 meters and has been crush-tested to 2,200. Full ocean depth, 6,000 meters, is far harder, and it is where he expects critical minerals to matter. The $50,000 figure is a base vehicle. The $10 trillion-plus figure is asserted, not derived.

His answer is structural. The smartest people in 1960 went into government. Now they go to San Francisco, so work once reserved for the state will move to private companies. "There is a future where Ulysses is the largest navy in the world." He casts himself as a heart-driven founder, in the Steve Jobs mould rather than the Jeff Bezos one, and credits faith as part of the motivation. Regulation is a separate drag. A seagrass project in Australia, possibly the biggest in the world, has stalled for a long time against what he calls green tape. Neither argument settles how many buyers exist. The factory due next year, built for a few hundred vehicles and maybe 1,000, will start to answer it.