Panthalassa, a renewable energy and ocean technology company, is building a new class of data center that floats in the deep ocean, powered by wave energy, and connected by Starlink satellites, with no transmission cable to shore.
Earlier this month they announced a $140 million Series B round, led by Peter Thiel, alongside Marc Benioff, John Doerr, and Max Levchin. The strength of investor interest has pushed the company’s valuation to almost $1 billion. The plan is to deploy Ocean-3, the company’s first commmercial node, in the northern Pacific later this year.
The investors are betting on it. The engineers and the marine scientists have questions. This is what each side is saying.
Why this matters for the blue economy.
Since the AI boom, the defining constraint on the industry has been electricity. Data centers consume enormous amounts of power, and building new grid capacity on land is slow, expensive, and increasingly contested by the communities that would host it. Companies are racing to find unconventional answers.
Wave energy is one of them. The premise is that wind drives waves, and the ocean carries that energy across thousands of miles before anything taps it. Panthalassa's strategy, in its own framing, is to put the computer where the energy already is, rather than move the energy to the computer.
A debate over whether the open ocean can actually power the AI industry is now unfolding: marquee investors on one side, energy analysts and marine scientists on the other.
What Panthalassa Is Building
Panthalassa is based in Oregon and was founded in 2016 as a public benefit corporation, a company structured to weigh public benefit alongside profit. It has spent close to a decade developing the pieces that have to work together: power generation, propulsion, autonomous operation, and computing.
The name is a clue to the ambition. Panthalassa was the single, worldwide ocean that surrounded the supercontinent Pangaea hundreds of millions of years ago, before today's separate seas existed. It is a fitting label for a company that wants to treat the open ocean as one continuous source of power.
Each Panthalassa node, per the company's public materials, is a roughly 85-meter steel column, about the height of Big Ben. It is built from earth-abundant materials, with no hinges, flaps, or gearboxes. Ocean water passes through a turbine to generate power. The compute itself sits in a sealed, seawater-cooled container that recirculates its working fluid internally. Nodes are fabricated at a pilot manufacturing facility in the United States, towed out horizontally, and rotated upright on site.
There is no electrical cable to shore. Connectivity runs through Starlink satellites. The product the company sells is not electricity; it is compute. If the design works, AI capacity stops competing with farms, towns, and substations for grid power and for land.
📸 credit: Panthalassa
What the Energy Analysts Are Asking
The central engineering question is whether Ocean-3 can survive a North Pacific winter.
Wave energy has a long record of clever designs that did not last. Pelamis, Aquamarine Power's Oyster, and Ocean Power Technologies are among the projects from the past two decades that ran into the same three forces: saltwater, storms, and stationkeeping.
Most failures came not from moving parts but from fatigue at the steel interfaces, corrosion at the joints, and biofouling. Barnacles, mussels, and algae eventually coat almost anything left in saltwater long enough to turn it into a reef.
Energy. A wave holds a lot of energy, but only some of it reaches the computer chip. Power is lost when the wave is caught, when it is turned into electricity, and when it runs through the wiring. Each step takes a slice. Independent observers note that public wave-energy cost data sits closer to $0.50 to $1.00 per kilowatt-hour than to the far lower figures sometimes cited in coverage of the sector.
Drift. A free-floating node is, by design, not anchored. A large permitted ocean area is therefore not a fixed site so much as a zone a node moves around within, which matters for maintenance and servicing.
Maintenance Economics. A current of half a meter per second can move an object roughly 26 miles per day. The practical challenge is also reliably locating, servicing, and maintaining it over time.
Insurance & Weather. Anything floating, autonomous, and expensive will eventually meet weather no model fully priced in. How underwriters treat a drifting compute fleet is an open question
Panthalassa's answer is straightforward. Simplicity at the unit level, including no gearboxes, no hinges, and earth-abundant materials, should reduces the failure modes that sank earlier projects. Whether that holds at fleet scale is exactly what Ocean-3 is designed to test.
The Ecological Questions
The ecological debate has barely started, but scrutiny is coming, and the regulatory framework already exists.
A 2,000-node drifting fleet would have to clear an environmental review. NOAA already has established frameworks for evaluating thermal discharge, plankton entrainment, and marine-mammal impacts.
Three issues stand out.
Thermal discharge and entrainment. Even small temperature changes around an intake or outflow can stress organisms already under heat stress from a warming ocean. NOAA's environmental assessment for a 400-megawatt ocean thermal energy plant estimated that on the order of 2,000 kilograms of plankton per day would be drawn into the warm-water intake. Microsoft's Project Natick, an underwater data center trial, is the precedent the company's supporters cite most: Microsoft reported the discharged water was a fraction of a degree above the surrounding ocean, with no harmful local effects observed. The caveat is that Natick was a single sealed vessel resting on the seabed for two years. A drifting fleet with active intake and outflow cycles is a different experiment. Panthalassa says its cooling loop recirculates internally, which could significantly reduce impacts. However, the company has not yet published thermal-discharge data.
Biofouling chemistry. Keeping heat exchangers clear in saltwater usually means antifouling chemistry. The literature on ocean thermal plants points to chlorine, bromine, and ozone as standard tools, and any of those, discharged at fleet scale, would need Clean Water Act review. Whether Panthalassa uses biocides, which ones, and in what volume is not yet in the public record. Earth-abundant materials answers the supply-chain question; it does not answer the chemistry question.
Marine mammals. NOAA Fisheries and the Bureau of Ocean Energy Management already maintain a joint framework for offshore wind and North Atlantic right whales, organized around four stressors: underwater noise, vessel strike, entanglement, and habitat displacement. All four apply to a fleet-scale autonomous compute array, and a drifting node array in the North Pacific would intersect known cetacean corridors. The towing-and-deployment phase, with vessels moving 85-meter structures over long distances, is the obvious early focal point. Major ocean-conservation groups have not weighed in publicly on Panthalassa yet. The likely trigger is a named deployment site or a draft environmental assessment. The silence is the early part of the story, not the absence of one.
What to Watch
Ocean-3. The first commercial node is due for deployment in the northern Pacific later this year. Whether it survives a full winter is the threshold test.
Site disclosure and operating data. A named deployment location would trigger federal review on thermal discharge, plankton entrainment, and cetacean impacts. Node-level metrics such as uptime, maintenance requirements, and delivered energy costs would move the discussion from theory to evidence.
Conservation response. Major ocean-conservation groups have not commented publicly. Once a site is named, they are likely to.
The first customer. A signed hyperscaler such as Google, Microsoft, or AWS, or a sovereign customer, would transform the story from a venture story into a revenue story.
The blue economy will absorb capital like this for the next decade, and the public debate over what the open ocean is for is starting now. Wave energy, offshore wind, ocean thermal power, carbon removal, and mariculture will all run into the same question: who gets to use the open ocean, and on what terms?
_____
Sources: Business Wire • CleanTechnica • SustainabilityMagazine • NOAA • CBS News
📸 Credit: Panthalassa
From the issue: Nel Blu Issue 002 (May 19, 2026)