back to article OpenAI to buy electricity from CEO Sam Altman's nuclear fusion side hustle

OpenAI is reportedly in talks with Helion to get access to the startup's not-yet-possible nuclear fusion-driven electricity generators. The negotiations were revealed by a Wall Street Journal report, which claims any deal would provide "vast quantities" of power to OpenAI for operating datacenters of machine-learning work. …

  1. Yet Another Anonymous coward Silver badge

    So in summary

    Not really a company invests money it doesn't have to build an imaginary supercomputer to do unspecified AI stuff using power from an imaginary reactor that's supposed to be built by a company that nobody believes can build it.

    Nice to know that some things ever change with AI.

    1. b0llchit Silver badge
      Coat

      Re: So in summary

      Please note: Hallucinated electrons are also both particles and waves. They are waves and waves of random information(*) until they hit you in the face electrifying you mood.

      (*) Unless you get to absolute zero, because at that temperature everything aligns in perfect harmony and harmonic truth will descent upon AI.

    2. Charlie Clark Silver badge

      Re: So in summary

      It's very close to fraud but investors will be pleased to be getting a piece of Helion at a perceived discount.

    3. Bebu
      Windows

      Re: So in summary

      Not really a company invests money it doesn't have to build an imaginary supercomputer to do unspecified AI stuff using power from an imaginary reactor that's supposed to be built by a company that nobody believes can build it.

      A level of virtualisation that even VMware haven't achieved. :)

      I guess everyone is hoping a quantum vacuum fluctuation wanders long dragging one of these imaginary steps into reality thereby bootstrapping the whole sorry fiction.

      Seems more like magical thinking which we should have abandoned after Newton's time. The Unseen University's wizards are far more practical than this circus and I am omitting High Energy Magic's Ponder Stibbens.

    4. JoeCool Silver badge

      Re: So in summary

      Can we call this "doing an Elon ?"

  2. Anonymous Coward
    Anonymous Coward

    Wait isn't nuclear fusion not really viable y-

    > the startup's not-yet-possible nuclear fusion

    Oh so it is just marketing.

  3. DS999 Silver badge

    I would love for Helion to succeed

    Not only fusion, but aneutronic direct capture fusion - that's the real holy grail.

    If AI can help make it work at least it will be doing something more useful than trying to figure out how to make someone click on Google Search ads 0.9% more often! If it does then we don't have to worry about how much power AI is using, because everyone will site a fusion reactor or two next door to their megadatacenter.

    1. IvyKing Bronze badge

      Re: I would love for Helion to succeed

      I would be nice if Helion could pull it off, but more than a half century of watching various fusion projects leaves me a bit skeptical.

      1. DS999 Silver badge

        Re: I would love for Helion to succeed

        Not saying I'm optimistic, but if Altman gets megarich off AI and plows more money into fusion than all the government sponsored research projects over that half century combined, who knows.

        1. Dinanziame Silver badge

          Re: I would love for Helion to succeed

          Admittedly, it seems fusion does not get a lot of funding compared to the benefits it's supposed to have. Maybe there's a catch that well-educated people don't talk about? Or maybe it's just really really hard and the research is too frustrating to attract a lot of interest.

          1. Filippo Silver badge

            Re: I would love for Helion to succeed

            Well, it gets significant funding. ITER is expensive. There are a few other fairly expensive projects. It could get more, surely, but it's not exactly starved. There's also the consideration that, so far, it looks like if and when making a fusion reactor that actually outputs power becomes feasible, it will be a loooong time before it can compete economically with other sources. It's a good thing to throw money at, but it's not a miracle.

            1. MiguelC Silver badge

              Re: I would love for Helion to succeed

              According to this article, ITER total costs might rise up to $65 billion, although ITER's management disputes this amount and maintains the official $22 billion estimate from 2016.

              1. sitta_europea Silver badge

                Re: I would love for Helion to succeed

                "According to this article, ITER total costs might rise up to $65 billion, although ITER's management disputes this amount and maintains the official $22 billion estimate from 2016."

                If we're going to talk about numbers with a broad brush, then let's look at the fossil fuel spends and try to put them into perspective.

                Currently, at grave danger to the survival of every living organism on the planet, even if ITER finally costs $65 billion over a few decades, we spend that much on fossil fuels EVERY THREE DAYS.

                https://en.wikipedia.org/wiki/World_energy_supply_and_consumption

                That we let the accountants run things while the planet is dying in front of our eyes is truly, monumentally, insane.

              2. Charlie Clark Silver badge

                Re: I would love for Helion to succeed

                And yet ITER still remains the closest to anything like a commercial product.

              3. MachDiamond Silver badge

                Re: I would love for Helion to succeed

                "According to this article, ITER total costs might rise up to $65 billion,"

                That cheap? California's HSR from Los Angeles to San Francisco is now estimated at over $100bn with a completion date about on the same horizon as when Fusion might be a thing.

            2. OrientalHero

              Re: I would love for Helion to succeed

              Today's investment vehicles aren't required to be immediately profitable - Amazon didn't post a profit for six years from startup (and if you believe their accounts in the UK as reported to the UK Tax Office, are only just profitable), Uber and the like come to mind too...

              1. jospanner Silver badge

                Re: I would love for Helion to succeed

                I do not believe their accounts in the UK as reported to the UK Tax Office.

              2. MachDiamond Silver badge

                Re: I would love for Helion to succeed

                d"Amazon didn't post a profit for six years from startup (and if you believe their accounts in the UK as reported to the UK Tax Office, are only just profitable)"

                The losses would be carried over for tax purposes and be deducted from net profits until they are gone. So when they first become net positive for a year, they weren't net positive over 7 years.

          2. CJ_C
            Facepalm

            Re: I would love for Helion to succeed

            Fusion has been only 25 years away, ever since I was in my 20s, 50 years ago...

        2. Charlie Clark Silver badge
          Facepalm

          Re: I would love for Helion to succeed

          Because he's rich, he must be clever?

          1. DS999 Silver badge

            Re: I would love for Helion to succeed

            Because he's rich, he must be clever?

            Where did I say that? You're just making shit up here, there is no way a reasonable person could have read my post to imply that. The only thing you could imply from my post is that the more rich he becomes the more money he has to potentially spend at Helion.

            So if he wants to drop billions on Helion to hire the best people and pursue a different path than ITER (one that, if successful, would result in a better product than ITER would end up with every possible aspect) then it doesn't matter if he's a genius or dumb as a stump, because he won't be the one in the lab tasked with making it work.

        3. Doctor Syntax Silver badge

          Re: I would love for Helion to succeed

          "if Altman gets megarich off AI and plows more money into fusion"

          But in order to get megarich he's going to need the cheap energy from the fusion reactor. To make this work he's rally going to need a complete virtual fusion reactor to spring into existence long enough for him to get rich - and who wants to be around when it springs out of existence again. Yes, I have just been re-reading In search of Schroedinger's Cat

      2. OrientalHero

        Re: I would love for Helion to succeed

        Agreed - almost 50 years ago I got to visit the JET in Oxfordshire, UK. They said then that Fusion power was 50 years away.

        Some companies now are saying it's only a decade away... I wonder what Helion has forecast and how realistic that is.

      3. Jaybus

        Re: I would love for Helion to succeed

        Well it is good to be skeptical, as long as you don't harbor a belief that it is impossible. Unlike many things that investors waste their money on that we don't actually know whether or not they are possible, we absolutely know that nuclear fusion is a reality. An AI that will discover new insights into the physical world may or may not be possible, but nuclear fusion is a known possibility. So, it is down to discovering a method that works, and sooner or later someone will discover it. Be skeptical of Helion, not of the possibility of fusion power. A half a century is but a blink of the eye in the timeline of paradigm shift events.

        1. MachDiamond Silver badge

          Re: I would love for Helion to succeed

          "we absolutely know that nuclear fusion is a reality."

          Yes and no. Obviously, the sun is a big fusion reactor with very low energy density and a thermo-nuclear bomb undergoes fusion very briefly. What isn't know is if fusion power generation is possible AND economic. So far, the emphasis has been on trying to get to the point of sustained fusion and on to continuously sustained and stable fusion. There there's the tricky bit of extracting that power without disturbing the reaction. There's also minor flourishes such as safety and security.

    2. Charlie Clark Silver badge
      Stop

      Re: I would love for Helion to succeed

      At the moment it looks to me like nothing more than another investor exercise in creative accounting. Big tech is already using certificates and promises from power generation schemes to make their own emissions look rosier even as they beef up their data centres.

      1. ecofeco Silver badge
        Pirate

        Re: I would love for Helion to succeed

        Because that's what it is.

        Smart people are starting to realize that big tech hasn't really been very innovative for a long time, just repacking the same old crap or straight up vaporware and "creative" accounting. Or just straight up price gouging and collusion.

        Not to mention obvious money laundering.

    3. Like a badger

      Re: I would love for Helion to succeed

      If it does then we don't have to worry about how much power AI is using, because everyone will site a fusion reactor or two next door to their megadatacenter.

      I doubt that. Just because it might work and fuel is expected to be abundant, it's still going to be unfeasibly expensive at first. Early iterations of cutting edge technology always tend to be astronomically expensive, and even production versions often turn out inordinately pricey. Back in 1954 Lewis Strauss, chairman of the US Atomic Energy Commission forecast that his children would benefit from atomic-electric power "too cheap to meter". In 1974 (and 1994, and 2014, and 2024) that's been proven wrong, and atomic power remains vastly expensive, such that it's very difficult indeed to create a viable commercial case.

      Fusion will be no different if it works, so any data processing outfit that think it'll be viable as a captive generator is kidding only themselves.

      1. Anonymous Coward
        Anonymous Coward

        Re: I would love for Helion to succeed

        Well the main reason for that with fission is the safety aspects of the reactor and its fuel, while its in use and the decommissioning of it. These have only gotten more expensive due to stricter requirements. Fussion shouldn't have these expensive running and decommissioning requirements.

        The reactor will likely cost vast sums to build at first, so will not provide it as cheap as renewable for a while, but would provide consistent, clean energy in any environment.

        The biggest problem with fussion is the abundant fuel, it's not abundant, He3 (aka helion) that most reactors are being designed to use is everywhere, but in very small quantities. In the atmosphere, it's calculated at parts per trillion, with a few thousand tonnes in the whole atmosphere. Natural gas contains it, but the total stock pile of in the US contains a few kgs. The amount of resources needed for the extraction and the environmental impact will be the biggest problem with most of the paths that are being taken to fussion at the moment.

      2. MachDiamond Silver badge

        Re: I would love for Helion to succeed

        "In 1974 (and 1994, and 2014, and 2024) that's been proven wrong, and atomic power remains vastly expensive, such that it's very difficult indeed to create a viable commercial case."

        Much of that is due to any innovation not being allowed. Alvin Weinberg, the father of the pressurized water reactor, saw very quickly that the PWR isn't a long term solution to nuclear power generation. The couple of commercial reactors of other designs didn't pan out and that's been it for decades. SMR's are a silly concept in my eyes. The anti-nuke protestors aren't displeased with the size of reactors, but their very existence in the first place so more and smaller reactors will only mean more lawsuits to prevent them going forward. For technical reasons, smaller PWR reactors are more difficult to operate due to neutron losses at the boundaries.

        The proponents of newer MSR designs have said that getting permissions is going to be more of a hill to climb than building prototypes. There's no way to get approvals to start building until there are approvals in place to allow operation. It's much safer for a government drone to say no to anything new. It's a poor idea to risk a very high paying job with loads of benefits by saying yes so something they don't have to.

    4. EricB123 Silver badge

      Re: I would love for Helion to succeed

      I'm looking into the economics of a small nuclear reactor in my basement.

  4. Anonymous Coward
    Anonymous Coward

    Fusion sans Altman

    I really really want fusion to be a reality but not with Sam. That guy gives me the creeps with his AI megalomania and doomsaying.

    Why do all these things need to come with a catch. In this case, Altman's involvement

  5. Pascal Monett Silver badge

    Fusion, fusion, fusion

    Funny how so many companies are talking AI, but hardly any are talking fusion. Must be because the budget is an order of magnitude higher for fusion and it's harder to bullshit your way through.

    Meanwhile, ITER is progressing. We will get there, apparently, but there is no billionnaire to thank for that. It would appear that space is an easier frontier for their low attention spans.

    1. Charlie Clark Silver badge

      Re: Fusion, fusion, fusion

      You must have been looking elsewhere because over the last few years fusion has attracted a lot of money. Many of them have some clever ideas but so far they've all slipped deadlines and seem to be facing the same problems in scaling up that ITER and the like have: plasma hot enough to fuse is hard too handle™. Which, incidentally is the title of my next book.

  6. Fruit and Nutcase Silver badge
    Alert

    Sam Altman's next acquisition

    A White Cat

    1. Bebu
      Pint

      Re: Sam Altman's next acquisition

      A white cat.

      Very good.

      Must be mandatory as even Danger Mouse's adversary, the toad Baron Silas Greenback, had Nero.

    2. Mike 137 Silver badge

      Re: Sam Altman's next acquisition

      Not just any old white cat -- it must also be extremely furry and tend to snarl ferociously.

    3. MachDiamond Silver badge

      Re: Sam Altman's next acquisition

      "A White Cat"

      They come with the Villain Starter Kit. Volcanic island lair sold separately.

  7. JoeCool Silver badge

    Not sure about this quote from the story

    "perhaps Helion aims to have more capacity than just that, and wants to sell its energy supply to OpenAI, which would need lots of power"

    Did you mean "Energy surplus"

    This strikes me as obsessing over a very remote (in time and probability) event; maybe Helion is desparate for R&D funds, and isn't so much worried about making the power numbers work, as it is about saying whatever is needed to get money.

  8. TimMaher Silver badge
    Mushroom

    Fusion. It’s going to take seven…

    “We can’t wait until next week!”

    … million years.

  9. b1k3rdude

    Er what could possibly go wrong with an unregulated body having access to nuclear power...

  10. Anonymous Coward
    Anonymous Coward

    Missing the boat?

    Should be good for power in another couple of decades.

  11. Daedalus

    There's a slight issue here

    Sure fusion power is more or less infinite in fuel supply, even if you're restricted to deuterium, but note the following:

    At the core of the Sun, energy is generated at a rate of less than 100W per cubic meter. For comparison your body does about 10 times that just sitting around. A compost heap does about the same without needing a temperature of 10 MK and a density thousands of times greater than water.

    An internal combustion engine generates power in the cylinders at roughly 50 MW per cubic meter, or 50 kW/litre.

    So your fusion technique needs to be 500,000 times better than the Sun to be comparable to fossil fuels. Or you could have a reaction chamber that's 100m square and 50m high.

    1. Filippo Silver badge

      Re: There's a slight issue here

      Nobody is trying to do fusion the same way it happens in the Sun.

      "The same source of power as the Sun" is a good slogan, but it's only true in the same sense that an internal combustion engine has the same source of power as a stove, because they're both combustion.

  12. Paul 195
    Mushroom

    Affordable fusion power

    We already have access to a working fusion power generator* that costs us nothing. And the technology to turn its output into usable electric current has been getting steadily cheaper every year for a while now. Although the idea of cheap limitless energy is appealing, it's still looking like quite a long shot; unlike most engineering problems it seems that scaling down is the problem not scaling up. The biggest problem with renewables is energy storage for when the sun isn't shining or the wind isn't blowing; what kind of excellent energy storage technologies could be developed for a fraction of the investment in chasing the fusion unicorn?

    * yes, I mean the sun.

    1. Filippo Silver badge

      Re: Affordable fusion power

      You are assuming that scientific research is like a factory, where you double the resources and you automatically get double the output. That's just not true. When you're doing research, paying twice as much is not going to halve time.

      Both fusion and storage are very well funded right now, with most interesting ideas being actively pursued already. Dropping one to focus on the other would result in complete loss of one avenue of research, while getting only marginal or even no benefit to the other. If either avenue has unforeseen side-benefits, which in science happens all the time, dropping one could even actively hurt the other.

      Stop pitting good ideas against each other to score points. The only smart solution is to do both. Which, fortunately, is exactly what we're doing.

    2. MachDiamond Silver badge

      Re: Affordable fusion power

      "We already have access to a working fusion power generator* that costs us nothing."

      At an effective maximum of 220W/M2, it's not dense enough for modern society at the efficiencies of cheap machinery. It's fine for my personal needs as I don't live in a high rise shoe box, but if I still had my manufacturing company, covering that roof (leased building, though) wouldn't not provide enough power and certainly wouldn't have had the required consistency. The business wasn't even one that could be called a heavy power user. Offices, soldering irons, curing ovens, that sort of thing.

    3. Doctor Syntax Silver badge

      Re: Affordable fusion power

      We've been harvesting energy from the sun for a very long time indeed. It's called growing food. One of the things about solar energy is that there's a strictly limited amount available at any one time and it has to be shared out. Currently we're seeing plans for solar farms, some fortunately being turned down by planners, presumably on the basis that we need electrical energy more than we need food.

  13. Anonymous Coward
    Anonymous Coward

    What the experts are saying

    Sounds like Acme Ai, Nuclear Reactors & Roller Skates, Inc. has some competition.

    -Wyle E Coyote, Acme Holdings CEO

    You are letting the Ai near the nuclear? Ha ha ha ha ha

    -Queen Borg, Board Member, Unimatrix Zero

    Fusion is for amateurs. I am working on a matter anti-matter collision device controlled by dilithium crystals that can unleash tremendous amounts of energy!

    -Zephram Cochran, Inventor

  14. ecofeco Silver badge
    Pirate

    And there it is

    When an employee sets up their own outside vendor, it is considered, at best, conflict of interest and at worst, fraud.

    When the CxOs do it, it's all quite legal.

    1. MachDiamond Silver badge

      Re: And there it is

      "When an employee sets up their own outside vendor, it is considered, at best, conflict of interest and at worst, fraud."

      That's a bit too general. If the idea was to locate in Iceland and install a geothermal power plant, not a big deal. OpenAI could benefit from a cheap and dedicated power supply and Sam would make a return on the investment. Fusion isn't a reality, so there's a risk it will never work or won't be viable in a reasonable time frame. Time is money.

  15. Anonymous Coward
    Boffin

    Virtual aneutronic nuclear fusion electricity generators

    Seems appropriate, a seller of virtual snakeoil is now involved in virtual electricity.

    Helion Energy .. are developing a magneto-inertial fusion technology to produce helium-3 and fusion power via aneutronic fusion”

  16. Zippy´s Sausage Factory
    Devil

    So they're throwing money at a technology that doesn't yet exist and nobody else has yet managed to work that just happens to have the same CEO?

    What are the chances of that happening? Must be millions to one, surely?

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like