Sam Altman took his turn at the center of the Davos whirl this year, and he didn’t just use his moment to talk about AI — he wanted to sell the world on nuclear fusion. “We need fusion,” the Open AI CEO told Bloomberg’s Brad Stone in an interview. “My whole model of the world is that the two important currencies of the future are compute/intelligence and energy,” he said. “There’s no way to get there without a breakthrough.” Altman joins a long list of celebrity tech CEOs who get dreamy about the long-awaited promise of fusion energy: Jeff Bezos, Bill Gates, Marc Benioff, Dustin Moskovitz, Peter Thiel. What is it about tech magnates and fusion? Psychologically, it’s not hard to see the appeal: Fusion is precisely the type of out-of-this-world technology that captures the attention of people who see themselves as moonshot-chasing disruptors. Fusion promises to turn the nuclear process that powers the sun and stars into virtually limitless amounts of energy with minimal environmental consequences. Of course, science still needs to solve some knotty problems — and as a result, fusion has been in the “exciting future promise” category for decades. (The field has a running joke: fusion power is just 30 years away, and it always will be.) Around the mid-2010s, Silicon Valley investors began financing new startups to explore various pathways to fusion. Many boasted that they could get to a working power plant much sooner than the scientific consensus. If you’re optimistic, the arrival of Valley money and can-do spirit is just what fusion needs. If you’ve been watching fusion and its struggles for decades, though, it might raise a few questions about the judgment of the tech moguls who look to it as a quick fix. “Silicon Valley got interested in the energy space and feel like — because they’re these masters of the universe that can do anything — that they can solve uniformly intractable problems on unholy superhuman timescales,” said Edwin Lyman, a physicist and director of nuclear power safety at the Union of Concerned Scientists. That mindset “may have met its match in trying to disrupt the careful engineering that’s necessary to develop a new power source,” he said. Then there’s the direct business interest. Altman himself is executive chairman of the nuclear fusion startup Helion Energy, into which he sank his biggest personal investment to date: $375 million in 2021. Last year, the company signed a deal with Microsoft to deliver it electricity from the world’s first fusion plant, scheduled to come online in 2028. (If that sounds unrealistic — the company told DFD the goal is “demanding,” but the “world’s growing energy needs aren’t waiting,” and it will incur financial penalties if the deadline is missed) From a PR perspective, fusion is perhaps even more appealing. The data centers that power ChatGPT and its rival next-generation AI language models are immense power hogs. Some early estimates predict that by 2027, the AI boom could devour as much electricity each year as an entire country — turning into a big liability for Big Tech companies that have been making net-zero emissions pledges. Any stable future for the AI business — economically and politically — will require access to a lot of power without being branded as the latest dirty industry. Altman’s solution to that problem is a radical change in our energy supply. Tech companies currently lead as the world’s biggest corporate buyers of renewable projects, but with AI specifically, there’s a reason fusion is more attractive than renewables like wind and solar: Like traditional power plants, it has the ability to provide the kind of consistent “baseload” energy that can power algorithms to run nonstop. Wind and solar, by contrast, vary by time of day, season and region — and require some kind of storage and transmission infrastructure. (It’s notable that Elon Musk, a battery mogul, is a skeptic of affordable fusion.) “You don’t have the luxury of turning [AI] off when the wind comes on and turning it on when the sun comes up,” said Egemen Kolemen, a Princeton University professor and staff physicist at DOE’s Princeton Plasma Physics Laboratory. “Fusion might be very useful. Not in a decade — I’m not that optimistic — but maybe in a few decades.” Is all the tech interest helping? Fusion startups have multiplied in recent years, as has the support after a 2022 national lab breakthrough finally created a fusion reaction that produced more energy than it consumed. Venture funding for nuclear also spiked 790 percent in 2021 when Russia’s invasion of Ukraine shocked global oil and gas markets. “The money is always good,” noted Kolemen, plus the accompanying hype has attracted talent and renewed urgency in the cause. The federal government seems to agree — awarding $46 million to eight startups last year, including the Gates and Bezos-backed Commonwealth Fusion Systems, for pursuing pilot plants within a decade. And there’s another attractive, if very speculative, vision for how AI and fusion can grow together: AI-powered tools could accelerate fusion research, the same way they’ve done for difficult problems in biology. Still, Carolyn Kuranz, professor of nuclear engineering at the University of Michigan, stresses that startups need to move toward peer-reviewed and open-source research — otherwise the field won’t benefit from much-needed steps forward. Given all the lavish future promises made for both technologies, there’s some chance we end up not far from where we are now. “If the future of AI depends on being able to commercialize nuclear fusion rapidly,” said Lyman, “then you’re essentially tethering one uncertain technology to another.”
| 
