Forgot your password?
Close
typodupeerror
AI Businesses The Almighty Buck

Silicon Valley Bets $200 Million On AI Data Centers Floating In the Ocean (arstechnica.com) 57

Posted by BeauHD from the what-to-expect dept.
An anonymous reader quotes a report from Ars Technica: Silicon Valley investors such as Palantir co-founder Peter Thiel have bet hundreds of millions of dollars on deploying AI data centers powered by waves in the middle of the world's oceans -- a move that coincides with tech companies facing mounting challenges in building AI data center projects on land. The latest investment round of $140 million is intended to help the company Panthalassa complete a pilot manufacturing facility near Portland, Oregon, and speed up deployments of wave-riding "nodes" designed to generate electrical power, according to a May 4 press release. Instead of sending renewable energy to a land-based data center, the floating nodes would directly power onboard AI chips and transmit inference tokens representing the AI models' outputs to customers worldwide via satellite link.

Each node resembles a huge steel sphere bobbing on the water with a tube-like structure extending vertically down beneath the surface. The wave motions drive water upward through the tube into a pressurized reservoir, where it can be released to spin a turbine generator that produces renewable energy for the AI chips on board. Panthalassa claims the node's AI chips would also get cooled using the surrounding water, which could offer another advantage over traditional data centers. "Ocean-based compute might offer a massive cooling advantage because the ambient temperature is so low," Lee said. "Land-based data centers use a lot of electricity and fresh water for cooling."

The newest node prototype, called Ocean-3, is scheduled for testing in the northern Pacific Ocean later in 2026. The latest version reaches about 85 meters in length and would stand nearly as tall as London's Big Ben or New York City's Flatiron Building, according to the Financial Times. Panthalassa has already tested several earlier prototypes of the wave energy converter technology, including the Ocean-1 in 2021 and the Ocean-2 that underwent a three-week sea trial off the coast of Washington state in February 2024. The company's CEO and co-founder, Garth Sheldon-Coulson, said in a CBS interview that he hopes to eventually deploy thousands of the nodes.
This discussion has been archived. No new comments can be posted.

Silicon Valley Bets $200 Million On AI Data Centers Floating In the Ocean More

Silicon Valley Bets $200 Million On AI Data Centers Floating In the Ocean

Comments Filter:

  • let me get this straight (Score:5, Insightful)

    by Thud457 ( 234763 ) on Wednesday May 06, 2026 @12:08PM (#66130484) Homepage Journal
    So we can't afford renewable wave energy for powering peoples' homes. But somehow it becomes feasible to power AI bros' desperate money grabs?

    • Re:let me get this straight (Score:5, Informative)

      by sg_oneill ( 159032 ) on Wednesday May 06, 2026 @12:41PM (#66130548)

      Undeniably the cheapest fuel sources are renewables now, wind, solar, and *probably* wave (I'm not sure if the maths done on that, but it seems plausible?)

      Instead we are now paying billions to NOT do wind power.

      This period of history is fueled by madness.

    • No, they can't afford the NIMBY movement. Yet.

    • Re: (Score:3)

      by AmiMoJo ( 196126 )

      Does anyone remember in Thundercats The Movie the baddies try to boil the sea? Probably where they got the idea from. It's okay because it's comically evil.

    • The answer was of course in the summary.

      Instead of sending renewable energy to a land-based data center, the floating nodes would directly power onboard AI chips and transmit inference tokens representing the AI models' outputs to customers worldwide via satellite link.

      Unless your home is floating on the node in the ocean the power has to be transmitted back to your land-based home.
      The transmission is the impractical part that this solution resolves by using the power at the node itself.

      No I am not new here, I just enjoy reading the summaries.

  • Easy way to kill this (Score:4, Funny)

    by thegarbz ( 1787294 ) on Wednesday May 06, 2026 @12:08PM (#66130486)

    Propose someone build one of these in the ocean opposite one of your king's golf courses. That'll end this stupidity.

    • Re: (Score:2)

      by ebunga ( 95613 )

      He's getting a cut, so he's all for it.

    • If these are located more than 11 miles offshore, are they not technically considered international waters? While there are jurisdiction issues, piracy is a concern. For wave and wind farms, there is very little for pirates to plunder unless they want to strip them down for raw materials like copper. A data center would have far more valuable and easier components to strip. Oil rigs have to worry about piracy but they are manned by a large crew; offshore data farms would be minimally staffed if at all.

      • Re: (Score:2)

        by Tablizer ( 95088 )

        Protecting one's assets probably would not look much different if invasion happens closer to shore versus further out. It's a vessel being protected, not water, per se.

        The only major difference I see is that within the boundary, the US Coast Guard would do the enforcing (barring something major), while the US Navy would if further out.

        • How many of these will the Coast Guard/Navy need to protect and how are they manned? With oil rigs, there is a large crew stationed to protect them some degree.

      • Re: (Score:2)

        by ebunga ( 95613 )

        They're within the exclusive economic zone, though.

    • Propose someone build one of these in the ocean opposite one of your king's golf courses. That'll end this stupidity.

      Seems like a great business strategy. Propose to build one offshore near one of King Trump's golf courses and then he'll pay you $1B of taxpayer money not to build it. Rinse, repeat.

  • They are not betting, they don't really loose when they loose. They are SPENDING $200 million on the idea. There is no potential for "failure" so it's not a bet.

  • Seems really low-maintenance to me (Score:4, Informative)

    by Locke2005 ( 849178 ) on Wednesday May 06, 2026 @12:16PM (#66130502)
    Steel structures in a salt water environment? What could possibly go wrong? (Anything with iron in it rusts very quickly at my beach house, including stainless and galvanized steel. The beach house was original build on... (wait for it!)... steel posts.
    • It will require Schrödinger levels amount of maintenance. It will be very little because AI can do everything. It will require a lot of maintenance because AI is incapable of doing real world things. Low maintenance makes the investment look good to suckers—I mean investors. Lots of maintenance means jobs to suckers—I mean the local population.

    • It only has to last for 4 years before the next model comes out

    • Re:Seems really low-maintenance to me (Score:4, Informative)

      by ceoyoyo ( 59147 ) on Wednesday May 06, 2026 @02:04PM (#66130714)

      Good point. We have absolutely no history of building steel structures in the ocean. Certainly not a fleet of them all around the world for the last century and a half.

      • Yeah, boat maintenance fucking sucks.

      • Good point. We have absolutely no history of building steel structures in the ocean. Certainly not a fleet of them all around the world for the last century and a half.

        Good point. We have absolutely no history of having sailors constantly painting those steel structures to keep them protected from the harsh environment. No sir. All of those steel structures are protected once and then forgotten about. No need for constant maintenance.

        (WTF)

        • Re: (Score:3)

          by ceoyoyo ( 59147 )

          You don't paint the immersed parts of ships unless they're at dry dock, which certainly doesn't happen "constantly." Pipelines, the immersed parts of offshore platforms, and other underwater structures are rarely if ever painted.

          Marine engineering has maintenance and other considerations, absolutely. It's definitely not the show stopper you imply. As I said, we do it all the time.

          (WTF)

          Right back at ya.

      • Yes, and those steel ships require constant maintenance, don't they? Especially the electrical systems.

        • Re: (Score:2)

          by ceoyoyo ( 59147 )

          They do not, no. They do require maintenance, as does every structure. These ones will too. We don't not build things because they require maintenance. We take into account the cost of it when deciding whether something is economical or not. And we do find that lots of steel structures in the ocean are economical to build.

    • Re: (Score:1)

      by Anonymous Coward

      Steel structures in a salt water environment? What could possibly go wrong?

      The operators of all the off-shore oil platforms might have some insight in this area. Undersea oil pipes, floating docks, lighthouses, bridge abutments, etc. all operate in a ocean environment for long durations. With proper maintenance, this is a solved problem.

    • And, the data center needs water for cooling, so that'll need a desalination plant to clean the water, and the water will have to get additives to inhibit corrosion and mineral build-up in the copper pipelines that run through the data center, so... polluting!

      Imagine if one gets stuck in the doldrums, or is it going to be anchored someplace that always has non-threatening waves? What about hurricanes and typhoons... how do they like that kind of waves?

    • A sea-surface heater that offers its victims compute cycles... The Venusians are diversifying in the face of solar and EV sales.

  • Why not use ocean-based wind turbines? (Score:4, Insightful)

    by PuddleBoy ( 544111 ) on Wednesday May 06, 2026 @12:52PM (#66130572)

    We have all these ocean-based wind turbine projects (that Trump is trying to kill) that sit in that cold, cold ocean already. Maybe you could site some underwater facilities at the base of them (and join them with other, nearby facilities and turbines via underwater cables) and you have a self-contained data center...

    Wind power is more developed than wave-based. Might cut costs...?

    • Re: (Score:2)

      by ukoda ( 537183 )
      Wave energy is more constant than wind. You can have windless days but waveless days are not really a thing, just how big the waves are. That said I like tidal generation because the moon and earths relative paths are stable and predictable hence when and how much power you can generate is easy to determine and the cycle is within the scale that batteries can cover the dead time when the tide is turning.

  • Seawater is a corrosive microbial soup, and I'm seeing that they will have it constantly circulating around through the entire structure. It will even be driving their turbines. An ecosystem of wildlife will grow in there and eventually clog it all up. I assume they've got it plated with biocides but that will eventually be defeated.

  • Let's just one-up the DCs in space and in the middle of the ocean to the middle of the planet. That is thinking too small. Why not? The Earth's center has unlimited DC energy, it is very secure (hard to get to), there are a few heat issues to work out and maybe some problems drilling but that's ok. Those are just some first principle issues to resolve. Let's start raising the trillions needed for the Middle Earth DC or MEDiC for short.
  • If America's powerful elites absolutely insist that there be no correlation between wealth and intelligence, then I would at least like them to agree in principle that the correlation shouldn't be quite so negative.

  • Sure, lets warm up the Pacific Ocean. What could go wrong ?

  • It wasn't cost effective.

  • Wrong headline again (Score:3)

    by ukoda ( 537183 ) on Wednesday May 06, 2026 @05:11PM (#66131000) Homepage
    Take that redundant and pointless data center and AI buzz words soup out the headline and the story is far more interesting. The linked video is well worth watching to see how this power generation system actually works. Wish them good luck with this power generation project, hopefully the tech won't be wasted on AI, and instead be used for something useful, such as replacing existing fossil fuel use.
  • Can't build a normal data center on land? Certainly putting computers into the corrosive ocean will be easier!

  • Floating power-gen is difficult (Score:3)

    by NeumannCons ( 798322 ) on Wednesday May 06, 2026 @09:47PM (#66131310) Homepage
    Wave power has been tried many many time and has been found at best to be really difficult. There's a lot of "in development" and ongoing studies. There have been several commercial launches and a lot of failures.

    These devices need maintenance - a lot. Marine life needs to be removed. Serviced frequently as saltwater is extremely corrosive. Strong storms can be incredibly destructive.

    MS tried the underwater DC (with power from shore) and found it had too many problems. Doubling down on difficult (lets have the DC gen its own power using largely unsuccessful waver power) does not make the overall project easier.

    Maybe they should add a 3rd new technology to really launch this. "All the underwater data centers will be serviced by modified Tesla Optimus robots wearing scuba gear and carrying harpoon guns to protect against pirates".
  • Beautifully put together video on the Panthalassa site, but the facts buried in it belie key assertions. They have created a gigantic expensive machine to accomplish what a $1000 diesel generator can provide. They are orders of magnitude off in their costs per kilowatt from being the least expensive energy source they purport they will be and I do not see how they are going to close that gap with scaled manufacturing. Giant devices that can survive in the ocean are expensive. They are building a 3D devi
    • It doesn't ever have to be cost effective, it just needs to look good enough on paper to attract suckers to invest in it, then skim off assets before quickly declaring bankruptcy... just like all the other data centers.

  • ...of sunk costs.

  • I hope these investors have done their homework. Salt water is a harsh mistress. Not crazy though. Step one: consult experienced ocean engineers. Build a bespoke ship that keeps salty, moist air out of the machine rooms. Run fiber cable to land and another cable to wave power generator(s). Aggressively and perpetually station-keep. Never ever lose power or run out of fuel. Maybe don't try to generate wave power onboard datacenter ship? Ships are good at floating and can run away if a big storm looms

Slashdot Top Deals

One possible reason that things aren't going according to plan is that there never was a plan in the first place.

Close