Unlocking the ocean economy

A trillion-dollar frontier in plain sight

(Welcome to the 1,000+ new Rational Optimists who joined this week! You’re about to read an ROS Deep Dive—we publish these deeper research pieces into important tech frontiers every 1-2 months.)

Five thousand meters deep in the Pacific Ocean, under pressure that would implode a military submarine, lies a vast field of trillions of rocks.

These potato-sized rocks are the richest known deposit of battery metals on Earth. They contain more nickel, manganese and cobalt than all mines on land combined.

And they’re just lying on the seabed of the Clarion Clipperton Zone (CCZ) between Hawaii and Mexico. Ready to be scooped up by whoever can solve the daunting engineering puzzle of retrieving anything from those depths economically.

For all the technological progress humans have made, we’re still pretty helpless in the oceans. Clueless too. We’ve mapped 100% of Mars but only 27% of the seafloor. Researchers are still discovering underwater mountains the size of Mount Fuji.

Imagine what else is down there?

For reasons you’ll learn in this month’s Deep Dive, ignorance of the oceans is no longer an option. Governments and militaries suddenly care a lot about the ocean. They’re urgently funding technologies to monitor, defend and harness its potential. An ecosystem of startups has sprouted to take on the challenge.

In deep tech investing circles, the ocean is hot right now.

Our friend Will O’Brien, founder of “SpaceX of the Sea” Ulysses, says the ocean is in the early stages of being “industrialized.” A trillion-dollar economy on the cusp of being unlocked. Similar to outer space, with less fanfare.

Let’s look at 10 fast-moving startups unlocking the new ocean economy. First, some background on what makes the ocean such an important and uniquely challenging frontier.

Harsh, huge and highly important

To understand humans’ relationship with the ocean, think of the three Hs.

First, the ocean is harsh to humans and our technologies. Saltwater is extremely corrosive. It eats through metal and destroys electronics.

The sea also thwarts our communication tech. GPS doesn’t work underwater. Starlink has been a game-changer and enabler of new technologies on the ocean, but it’s useless beneath the waves.

Radio signals can only reach about 20 meters into water. Acoustic signals can travel underwater but are sluggish.

Then there’s the pressure. If we dive too deep or too fast, the ocean crushes our vessels. Then there’s “biofouling,” which Anduril’s Shane Arnott (more on Anduril in a bit) emphasized is a big challenge with autonomous submarines. Organisms colonize anything you put in the water, dragging on subs and blinding their sensors.

In short, the ocean presents a big new problem set for entrepreneurs to solve.

Second, the ocean is huge. It covers about 360 million square kilometers, more than twice as big as all land. But that’s just the surface. The ocean is 3-dimensional, comprising 99% of all habitable space on earth, with an average depth of around 4 kilometers. To really know what’s going on down there, we’ll need insight into each layer.

Third, the ocean is highly important.

In February 2024, a ship called the Rubymar took a Houthi missile to its hull in the Red Sea. The crew abandoned ship, dropping anchor as they fled. The ship would drift for two weeks, its anchor scraping along the bottom.

The anchor severed three undersea cables, knocking out roughly a quarter of the internet traffic running between Europe and Asia.

The ocean floor is humanity’s biggest blind spot. It’s also where our arguably most important infrastructure is. Ninety-five percent of all intercontinental data travels through roughly 500 fiber-optic cables resting on the seabed.

Your emails, bank transfers and favorite AI model all ride on these glass threads that transmit $10 trillion in financial transactions every day.

If enough of them were cut, the global economy would seize. Someone is cutting them.

In 2024 several cables were likely sabotaged in the Baltic Sea. Taiwan’s cables have been cut nearly 30 times in recent years. China just unveiled a deep-sea robotic arm capable of cutting cables at 4,000 meters, a depth where no human diver can repair them.

Meanwhile there are only 22 cable-repair ships on Earth.

There are also over 20,000 miles of active oil and gas pipelines snaking the seabed. We cannot even monitor them, let alone defend them. We still don’t know for sure who sabotaged the Nord Stream pipelines.

Militaries are racing to develop new underwater technologies. Late last year, a Ukrainian underwater drone named “Subsea Baby” heavily damaged a $400 million Russian submarine. The submarine was in the Russian port of Novorossiysk, supposedly safe behind layers of concrete barriers and floating booms.

Ukraine has effectively no navy. Yet it has forced the world’s third-largest Navy to retreat from its own home base in the Black Sea. Ukraine “hacked” a robot navy together, taking what were essentially jet skis and packing them with explosives. Swarms of these $200,000 drone boats hunt in packs, disabling warships and even shooting down fighter jets.

This tells us everything about where the ocean is headed. Just as drones have transformed warfare in the skies, drone boats are doing the same in the water. The same autonomous systems reshaping naval warfare are also reshaping commercial shipping.

Like drones, autonomous boats will birth industries we haven’t even imagined yet.

There is so much we can’t do in the ocean today simply because it’s too huge, harsh and expensive. Let’s meet the startups changing that. We’ll start in the sky above the ocean, then work our way down to the seafloor.

Sky

Poseidon Aerospace Autonomous seaplanes that use water as a runway

The first thing we noticed in Poseidon’s warehouse is the biggest American flag we’ve ever seen. The second thing we noticed is how fast they’re moving.

We met David Zagaynov, founder of Poseidon, for the third time last month in San Francisco’s Dogpatch neighborhood. On our first visit in April 2025 there wasn’t much to see. Just founder David and his little 13-foot-wingspan unmanned plane all alone in a giant warehouse.

When we returned only one year later, the factory was full to the brim. Three huge aircraft were being built.

Poseidon has reimagined cargo planes. Every cargo plane today was designed to carry people. The planes flying your Amazon packages are retrofitted passenger aircraft. They’re designed around things humans need—cockpits, windows, pressurized cabins.

Once humans are removed, you don’t need any of that.

Remove the pilot and you can get rid of the cockpit. No cockpit means no windows. No windows mean no pressurized cabin. No pressurization means a much lighter airframe made from carbon fiber.

A lighter airframe means smaller, cheaper engines. Smaller engines mean less fuel. Less fuel means lower operating costs.

Poseidon is targeting $400 to $500 per flight hour, which is about half the cost of regional carriers.

It’s now designing two 50-foot-wingspan planes. Heron is the big seaplane. It takes off and lands on water, with no need for a runway or port. It connects via Starlink and can fly autonomously, with remote human pilots able to take control if needed. Heron is targeting flight tests mid-2026.

Heron will carry cargo for 60% cheaper than regional cargo aircraft. David’s goal is to put all the regional cargo airlines out of business. With numbers like that, he just might succeed.

Egret is Heron’s twin but takes off and lands on runways. Same wings, same tail, same avionics, same engines—different fuselage, optimized for short takeoff and landing on rough strips.

Poseidon has also attracted serious interest from the military. Destroying your enemy’s runways has always been key to crippling their air power. Autonomous planes that can take off from any body of water change that.

Standing in Poseidon’s warehouse, we listened to engine test rigs roar while three carbon-fiber aircraft took shape simultaneously. We’re enamored by great ideas as much as anyone. But we’ve found the #1 indicator of success is speed of execution.

This company is about to get very big, very fast.

Surface

Saildrone Building a permanent ocean presence

Most ocean surveillance works like this: Send a ship. It burns diesel. A crew of over 20 stares at screens for a few weeks. Then everyone goes home. The ocean goes dark again.

Saildrone is pioneering “always on” sea surveillance.

We saw one of Saildrone’s boats near the Naval Air Station Alameda outside of SF. It looks like a bright orange sailboard, about 20 feet long. A tall carbon-fiber wing sticks straight up like a dorsal fin. It moves using wind and “thinks” using solar-powered computers. It can stay at sea for over 12 months.

Already over 100 saildrones are scattered across the world’s oceans. They quietly collect data that used to require billion-dollar research fleets. Saildrone’s boats measure the ocean from the surface down in a column and beam it all back via satellite in real time.

Saildrone has partnered with Meta to map the seafloor along a potential internet cable path. It used to take a traditional survey ship up to $200,000 per day to do this. Now all it takes is a small robo-sailboat.

Saildrone is the definition of “dual use.” The same drone platform that tracks illegal fishing fleets can hunt submarines for the US Navy.

The fleet has now sailed millions of nautical miles across every ocean. NOAA, the US Navy, the Coast Guard, NASA and dozens of commercial customers use them.

Saronic Mass-produces small, cheap, autonomous boats

Last year America built five oceangoing ships. China built over 1,000. One Chinese shipyard now makes more ships than the entire US maritime industry combined.

Austin-based Saronic is building fleets of autonomous boats that roll off a production line like cars. It’s the third most valuable defense startup in America, behind only Anduril and autonomous drone leader Shield AI.

Saronic’s approach reminds us of SpaceX. Before Elon Musk launched skyscraper-sized rockets, he built Starhopper, a squat little capsule that proved reusability worked.

Saronic’s Spyglass is the Starhopper of the sea. It’s a 6-foot stealthy reconnaissance drone boat—small enough to launch from another vessel at sea. It proved the core autonomy stack works.

Next up is Corsair, a 24-foot gunboat with a 1,000-nautical-mile range and a top speed over 35 knots. Saronic just landed a $392 million Navy contract for Corsair.

Marauder is the big daddy. It’s a 180-foot, 40-ton autonomous warship that can sail for more than a month without a human on board.

Each vessel in the lineup is designed to operate alone or in coordinated swarms. A single human operator will be able to manage dozens of boats remotely.

Saronic isn’t just building boats. It’s building the factory that builds the boats. It recently announced a $300 million investment to expand Franklin Shipyard in Louisiana, expected to come online in early 2027.

This is only the start. Saronic plans to build a mega-shipyard called Port Alpha. It will have 10x the production capacity of the Louisiana shipyard. Saronic aims to build it into the shipyard of the future—a factory that produces autonomous boats with the speed and consistency of a Tesla Gigafactory. No location has been announced yet.

Saronic’s mission is to flip asymmetric warfare back in America’s favor. The US has the most advanced submarines in the world, but each one costs billions of dollars and takes years to build. China has anti-ship ballistic missiles designed to sink carriers from a thousand miles away. Lose one, and it’s a national catastrophe.

Now imagine an enemy’s dilemma when facing a swarm of Saronic boats. Do you waste a $2 million missile to sink a far less expensive robot boat? What about the 50 boats after that?

As with aerial drones, the side with only expensive weapons goes broke.

Blue Water Autonomy

Toyota of the sea

During WWII America built 2,710 Liberty Ships in four years. They were ugly, welded cargo haulers that helped win the war because there were so damn many of them.

Blue Water’s approach is the opposite of Saronic’s. While Saronic is designing new autonomous vessels from scratch, Blue Water is taking a proven hull design and re-engineering it for autonomy.

Its flagship is the Liberty-class Unmanned Surface Vessel, a 60-meter steel autonomous warship with a 10,000-nautical-mile range. It will be able to sail continuously for three months without a human on board.

Blue Water gutted everything inside the ship that was designed for humans. Automated control systems and modular payload bays took their place. This simplifies the boat, making it cheaper to produce. The boats can be configured for reconnaissance, mine countermeasures, or whatever the mission demands.

Construction started on its first vessel in March. It’s expected to be delivered to the US Navy before the end of this year. If Blue Water achieves that, it will have gone from its founding to putting a ship in the water in two years with only $64 million. That’s the type of “Henry Kaiser” building we need. The Navy’s newest Ford-class aircraft carrier took 13 years and $13 billion.

Importantly, Blue Water’s ships are designed to be built quickly in various US shipyards, rather than highly specialized naval yards. By taking hull designs that already exist and modifying them, Blue Water is moving very fast. Like Toyota, Blue Water prioritizes mass-producibility, reliability and speed.

Underwater

Anduril

Building autonomous submarines

The deep ocean is the last militarily “uncontested domain.” That’s what Shane Arnott, Senior Vice President of Programs and Engineering at Anduril’s Maritime Division, told us at Anduril’s HQ in Costa Mesa last month.

The sky has drones and counter-drones. Land is surveilled from orbit down to the meter. Even outer space is now considered a “warfighting domain” by the US Space Force.

But underwater is… practically empty. Whichever country gets there first will secure a huge strategic advantage.

Arnott runs Anduril’s underwater portfolio. He’s building autonomous tools to dominate the ocean. Like Ghost Shark, which is roughly the size of a school bus. It’s an autonomous submarine that will be able to operate for months at sea at depths that would kill human crewed subs. It can hunt enemy submarines and map minefields. We could use a few Ghost Sharks in the Strait of Hormuz.

Ghost Shark’s design is radically different from a traditional submarine. A conventional sub keeps its hull pressurized and dry for the humans inside. Ghost Shark is “flooded”—seawater freely enters the hull. This makes it cheaper, simpler and capable of diving to previously unreachable depths.

(Notice a theme between Poseidon’s planes, Saronic’s boats and Anduril’s subs? They’re all designed for autonomy. Taking humans out of the equation lets you delete parts, which opens many new opportunities.)

The first Ghost Shark was delivered to the Royal Australian Navy in January 2026.

Anduril also makes Dive-LD, a large autonomous underwater vehicle built for intelligence, surveillance and reconnaissance. It’s already delivered to the US Navy.

Your authors with Shane Arnott. Ghost Shark is on the left; Dive LD is on the right.

Anduril’s maritime division is its only business with a civilian revenue stream. Its autonomous subs inspect subsea cables and map the ocean floor for commercial clients.

Achieving autonomy undersea is much harder than on land or in the air. There’s no GPS or Wi-Fi. Radio waves die in water. Communications bandwidth is almost zero. Arnott told us 70% of the ocean floor is essentially guesswork, even in classified databases.

Anduril is beginning to get underwater autonomy and communications to work by what Arnott calls “clever tricks.”

When we pressed for detail, he said Anduril is finding techniques that were developed decades ago and locked away in classified programs. Novel inertial navigation. Terrain-relative positioning. The best innovations in defense, Arnott says, are often old secrets repackaged.

We also asked Arnott what’s the most important thing nobody is paying attention to. He didn’t hesitate: the Arctic. The melting ice is opening new shipping routes and exposing vast mineral resources. China, Russia, the US and the UK are all positioning.

Ulysses SpaceX of the sea

The Colombian Navy recently seized an unmanned narco-sub. It was fitted with a Starlink dish, twin surveillance cameras and enough room to carry 1.5 tons of cocaine over 800 miles.

For every drug boat seized, you must assume 10 slipped through. The ocean is simply too vast to police with human crews. There aren’t enough ships, sailors or hours in the day to watch 140 million square miles of water.

The vibe at Ulysses is pure hard-tech startup. Tools everywhere. Parts scattered on benches. Carbon-fiber prototypes in various states of assembly.

Ulysses started with the beautiful mission of healing the ocean. Its underwater drones plant seagrass, which absorbs carbon 35 times more efficiently than a rainforest. In Australia, Ulysses cut restoration costs by 90% compared to human divers and replanted seagrass 10 times faster. Robo-gardeners for the sea.

Now Ulysses is building a Swiss Army knife for the ocean: a modular, autonomous system that can do almost anything underwater.

On the surface sits Leviathan, an autonomous mothership with hybrid-electric propulsion. Leviathan carries up to four Mako drones—small autonomous underwater vehicles rated to 1,500 meters. Makos are designed to go anywhere and do anything. They can plant seagrass on Monday then inspect a subsea cable on Tuesday.

A launch-and-recovery system called Kraken handles deployment and recharging autonomously. No humans required at any stage of the mission.

Will told us his goal is to put millions of these drones in the water, available for anyone to rent via an app. “Underwater work is too dangerous, costly, and difficult to do manually,” Will told us. “So much of it simply doesn’t get done today.”

It’s not just that Ulysses makes existing underwater work cheaper. It’s that an entire category of work—continuous monitoring, real-time seafloor awareness, persistent subsea presence—has never been done because it was economically impossible.

Ulysses is making it possible, just as SpaceX did for space. Just last month a16z led Ulysses’ Series A funding round, investing $38 million. Congrats Will and team!

Ocean floor

Standard Subsea

Inspection as a service

The world spends roughly $200 billion per year installing underwater infrastructure. Then another $110 billion per year maintaining it.

The way we inspect these valuable assets is comically inefficient. Hire a big crewed vessel, which costs up to $200,000 per day. Send a team out with a tethered robot. They inspect one stretch of cable or pipeline, pack up, and sail home. The infrastructure goes unchecked again for a year or two.

Roughly 200 cable faults occur per year globally. Average repair cost: $1 million to $3 million per incident. When something breaks it can take over a month until help arrives.

Standard Subsea founder Chris Amidon aims to transform this market. Its first product is called Scout. It’s an 18-foot surface vessel with hybrid-electric diesel propulsion. It looks like a normal boat but can be remotely piloted to an inspection site.

Scout then deploys a tethered remotely operated underwater drone. It descends up to 1,000 feet, capturing crystal-clear 4K video and sonar readings. When the job is done, Scout recovers the drone and sails home.

Standard Subsea’s business model is “inspection as a service.” It owns and operates the fleet, selling services on a subscription basis.

We asked Chris, “Why not make your boats autonomous?” He said it would only slow them down. He has such a huge cost advantage over any competition, it’s a drop in the bucket to pay a human operator to pilot the boat. Autonomy may come. For now, speed’s the goal.

The Metals Company

Mining the seabed (or trying to)

Remember those battery-metal-rich rocks sitting on the seafloor in the CCZ?

The Metals Company (TMC), working with its partner Allseas, is the only company in the modern era to successfully extract a load from the CCZ. It’s also the only publicly-traded company of the ten covered here.

Retrieving rocks from 4,000+ meters deep is extremely difficult. Pressure at those depths crushes most equipment. It’s pitch-black and freezing cold. And there are serious environmental concerns about kicking up plumes of debris that disturb the fragile, untouched ecosystems.

TMC has solved several of those challenges. A ship on the surface lowers a “riser system” in the water. It’s basically a 4-kilometer-long pipe that reaches down to the seabed. A robotic collector vehicle then grabs rocks and pumps them 4 kilometers up to the ship.

During pilot trials, it successfully recovered over 3,000 tons from depths of 4,300 meters. It was able to minimize disturbance with its proprietary Hopper 2.0 that captures 95% to 98% of the sediment and gently lays it back down.

This is not good enough for environmental groups, who oppose deep-sea mining on principle. Greenpeace calls TMC’s mission the Plunder of the Pacific. This is the same Greenpeace that still proudly opposes nuclear power in 2026. Factual analysis is not their strong suit.

Besides, everything’s a tradeoff. The world needs staggering amounts of nickel, cobalt and manganese to build the batteries that power electric vehicles and grid storage. Today many of these resources are mined by hand in dangerous conditions—like cobalt in the Congo—or controlled by China.

The ocean floor may be the West’s chance to break this reliance. The US President recently signed an executive order clearing a path for US companies to mine international waters. Deep-sea mining is about to go from curiosity to one of the most contested frontiers on Earth.

We are rooting for TMC, but this is a brutal business. The company is pre-revenue. We would not suggest buying its stock with retirement money.

Impossible Metals also deserves a mention. It’s building an autonomous underwater robot named Eureka that selects and picks up only high-value individual rocks.

Data and communications

AndrenaM

Building a distributed acoustic network to make the ocean transparent

Here’s how we “listen” to the ocean today: A person sits in a dark room wearing headphones, staring at green lines on a screen, trying to distinguish an enemy submarine from a shrimp colony. It is painstaking, highly skilled work.

These folks are called sonar technicians. There aren’t nearly enough of them to monitor 140 million square miles of ocean.

AndrenaM is applying AI to figure out what’s under the water.

It makes small, semi-autonomous buoys that sit on the ocean surface, each equipped with underwater microphones and onboard processors. The buoys can hear everything from the hum of a submarine to the ping of a fish finder. Each buoy performs acoustic analysis before sending the data via satellite.

Then AI models classify, track and identify every sound in the water. The AI cross-references acoustic signatures with the tracking transponders ships are required to carry. When a ship turns off its transponder, as military vessels and smugglers often do, AndrenaM keeps tracking it.

AndrenaM envisions thousands of networked buoys feeding into a single AI brain. The market AndrenaM is chasing is enormous and growing rapidly thanks to everything else you’ve read in this Deep Dive.

CSignum

Pioneering underwater Wi-Fi

We cannot wirelessly send data from underwater to above the surface.

This means every transmitter bolted to a pipeline has no way to talk to the world above unless at some point another device, usually involving an expensive cable, carries its signal through the surface.

This is why the ocean has a data problem. Everything below the water is “offline.”

CSignum, a small company out of Bathgate, Scotland, seems to have cracked it.

It uses low-frequency electromagnetic field signaling to carry data wirelessly through water to receivers above the surface. Whereas radio waves are absorbed almost instantly by water, these lower-frequency fields can push through.

It can transfer data up to 170 meters and up to 2 kilometers when paired with an acoustic gateway.

The product is already deployed with customers in the UK, EU and US.

There are dozens of use cases for wireless below-to-above surface connectivity.

For one, we can mount sensors to the bottom of ships and beam data to the cabin without wires.

Another: today, if you want to monitor an undersea cable, you either install expensive sensors tethered by cables or hire a ship of people.

Using CSignum’s below-to-above surface wireless technology, we could simply drop a box in the ocean for a fraction of the cost.

Scale this to the whole ocean, and we’ll be able to monitor every coral reef, undersea cable and offshore wind turbine.

Underwater data centers

There’s a lot of hype around launching data centers into space. Did you know we’re also experimenting with putting them on the seabed?

In 2018 Microsoft lowered a steel cylinder packed with 864 servers to the seafloor off Scotland’s Orkney Islands. The container was connected to shore by a single fiber-optic cable.

When Microsoft hauled it back up two years later, they found a surprise. The underwater servers experienced eight times fewer hardware failures than data centers on land.

Data centers have a heat problem. Stanford has found that 25% to 40% of a typical facility’s total electricity is spent cooling the chips. Submerge a sealed container of servers in cold seawater, and cooling becomes nearly free.

Microsoft shelved Project Natick. But others are running with it. China has deployed the world’s first commercial underwater data center cluster in the South China Sea.

US startup Subsea Cloud is deploying submarine data center pods near Port Angeles, Washington.

The cables that power “the cloud” are already underwater. Will computers follow?

The sea is becoming autonomous

To sum up what’s happening in the ocean in 4 words:

Autonomy is unlocking everything.

Isn’t it wild to think we can’t monitor our most important infrastructure? Or that Iran was unable to find its own mines it laid in the Strait of Hormuz, according to the NYT?

The ocean is far too large for humans. We need autonomous technology to unlock its potential.

That’s now happening thanks to “ideas having sex,” to steal a line from Matt Ridley. Three technologies have converged to let us build vessels that think for themselves and update like iPhones:

Communications. Starlink satellites in low Earth orbit have turned the ocean from a disconnected void into a networked space.

Energy. Cheaper, denser batteries enable a new class of electric underwater vehicles.

Code. Old shipbuilding was about welding steel plates. New shipbuilding is about wrapping a hull around a computer.

It’s an exciting time for the ocean.

Although we’ve only explored a fraction of it, the ocean has already given us several breakthroughs. Paraphrasing Ulysses founder Will O’Brien, writing for the excellent Not Boring publication:

Marine organisms have given us:

• Ziconotide - a painkiller 1,000x more potent than morphine, from cone snail venom

• Trabectedin - a cancer drug, from a sea squirt

• The green fluorescent protein (GFP) that won the 2008 Nobel Prize in Chemistry. It “enables scientists to track, amongst other things, how cancer tumors form new blood vessels, how Alzheimer’s disease kills brain neurons and how HIV-infected cells produce new viruses.”

Imagine what else is waiting to be discovered down there?

Let’s find out.

—Dan Steinhart and Stephen McBride