 |
Profile on: HELION: THE OPEN AI OF FUSION? Can the consistently overpromising nuclear-fusion company really deliver by 2028? By Berit Anderson Why Read: Helion has promised Microsoft a working nuclear-fusion reactor by 2028. Is it building state-of-the-art energy technology, poised to overturn energy markets, or another overleveraged, Sam Altman-type hype machine? ______
LLMs and Fusion: A Match Made in VC Hypeland You may not think so at first glance, but the business of nuclear fusion and LLMs have a lot in common. Both technologies cost billions of dollars to advance and are complicated enough that the average person or *cough cough* investor has no idea if the specifics of any given approach will really work. One is touted as the energy source of the future; the other, the general artificial intelligence of the future. Both still have a long road ahead to meet these lofty goals. Strategically, their futures are closely intertwined, as cloud providers, model creators, and chipmakers grasp at non-grid-connected energy sources they can use to stuff their gaping, power-hungry maws. Microsoft is betting heavily on both, having signed a 2023 purchase agreement with fusion company Helion Energy for a plant to be online by 2028 and generating 50 MW+ after a one-year ramp-up period. Reportedly, if Helion doesn't meet that goal, there will be financial penalties - although exactly what type doesn't seem to be a matter of public record yet. "The planned operational date for this first of its kind facility is significantly sooner than typical projections for deployment of commercial fusion power," a Helion press release boasted at the time of the deal. There's just one potential problem here: Helion has a long history of overpromising and underdelivering. "Here are some claims that have been made in the last 10 Years," Redditor blurpsn posted a full two years ago, already then among a growing chorus of Helion critics: "The Helion Fusion Engine will enable profitable fusion energy in 2019." - NBF [Nextbigfuture.com] 7/18/2014 "If our physics holds, we hope to reach that goal (net energy gain) in the next three years." - D. Kirtley, CEO of Helion in the Wall Street Journal 2014 "Helion will demonstrate net energy gain within 24 months, and 50-MWe pilot plant by 2019." - NBF 8/18/2015 "Helion will attain net energy output within a couple of years and commercial power in 6 years." - Science News 1/27/2016 "Helion plans to reach break-even energy generation in less than three years, nearly ten times faster than ITER." - NBF 10/1/2018 Their newest claim on their website is: "We expect that Polaris will be able to demonstrate the production of a small amount of net electricity by 2024." Dear reader: There was not net electricity by 2024. There was, however, significant additional hype and nearly half a billion in additional funding. Sam Altman threw in $375M of his own capital back in 2021, calling it the biggest personal investment he'd ever made. (Ah, if only we could go back in time and warn him about his future.) Just a few weeks ago, now that OpenAI's entire strategy relies on the ability to generate cheap energy, he made a follow-on investment in Helion's $475M Series F, joining OpenAI funder Softbank's Vision 2 Fund.
Fusion for Non-Nuclear Scientists |
