TechCrunch Desktop Logo
A fusion reactor glows purple.
Image Credits:Helion
Climate

Sam Altman-backed fusion startup Helion in talks to sell power to OpenAI

Tim De Chant

OpenAI CEO Sam Altman is stepping down as board chair of the Helion — the fusion startup he backs — amid reported talks between the two companies.

The deal, which was reported by Axios, is in early stages, and it could guarantee OpenAI 12.5% of Helion’s production — five gigawatts by 2030 and 50 gigawatts by 2035. OpenAI partner Microsoft signed a similar deal with Helion in 2023 to buy power starting in 2028.

If the figures in Axios’ report prove to be accurate, it suggests that Helion expects to be able to rapidly scale production of its fusion power plant. The startup has said that each of its reactors will generate 50 megawatts of electricity, meaning it will need to build and install 800 reactors by 2030 and an additional 7,200 by 2035. 

Helion wouldn’t confirm if talks with OpenAI were underway. A spokesman told TechCrunch the company has not announced any new customer agreements beyond those it already has with Microsoft and Nucor. However, the company did confirm to TechCrunch that Altman is leaving the board chair of Helion, suggesting that the two companies may eventually work together.

“Sam is stepping down from Helion’s Board of Directors after more than a decade. This decision enables Helion and OpenAI to partner on future opportunities to bring zero-carbon, safe electricity to the world,” David Kirtley, co-founder and CEO of the company, told TechCrunch in statement. “We look forward to continuing to work with him in this new capacity.”

Helion is racing to build its first commercial-scale reactor by that time. If the startup is successful, it would place it years ahead of the competition, which is mostly targeting early 2030s for commercial operations. 

The startup raised $425 million last year from investors, including Altman as well as firms Mithril, Lightspeed, and SoftBank.

Most fusion startups are pursuing one of two approaches — harvesting heat from the fusion reactions and using a steam turbine to turn it into electricity. Helion is taking a different tack, developing a reactor design that would use magnets to convert fusion energy into electricity.

0 of 31 minutes, 2 secondsVolume 0%
00:02
31:00
31:02
 

Inside the hourglass-shaped reactor, fusion fuel is first turned into plasma at either end and then shot toward each other using magnetic fields. When they collide in the middle, another set of magnets compresses the merged plasma ball until fusion occurs. The reaction pushes back on the magnets, which can convert that energy directly into electricity.

Helion is currently operating its Polaris prototype in advance of its push to commercial power. In February, the company generated plasmas inside the reactor that hit 150 million degrees Celsius, almost to the 200 million degrees Celsius the company thinks will be required for commercial operations.

Though Altman has stepped down from his position as chair of Helion’s board and reportedly recused himself from the discussions, his fingerprints are all over the matchmaking. 

Last year, Altman stepped down as board chair of Oklo, a small modular nuclear reactor startup that had merged with his acquisition company, AltC. The move was intended to allow Oklo to explore strategic partnerships with leading AI companies, including potentially with OpenAI,” Caroline Cochran, Oklo’s co-founder and chief operating officer, said in a statement given to CNBC at the time.

Update 1:30 pm ET: Added confirmation from Helion regarding Altman stepping down as board chair.

Topics

, , , , , , ,

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Tim De Chant

Tim De Chant is a senior climate reporter at TechCrunch. He has written for a wide range of publications, including Wired magazine, the Chicago Tribune, Ars Technica, The Wire China, and NOVA Next, where he was founding editor.

De Chant is also a lecturer in MIT’s Graduate Program in Science Writing, and he was awarded a Knight Science Journalism Fellowship at MIT in 2018, during which time he studied climate technologies and explored new business models for journalism. He received his PhD in environmental science, policy, and management from the University of California, Berkeley, and his BA degree in environmental studies, English, and biology from St. Olaf College.

You can contact or verify outreach from Tim by emailing tim.dechant@techcrunch.com.

View Bio
May 27
Athens, Greece


StrictlyVC Athens is up next. Hear unfiltered insights straight from Europe’s tech leaders and connect with the people shaping what’s ahead. Lock in your spot before it’s gone.

REGISTER NOW
Keep reading
Image Credits:Benjamin Fanjoy / Getty Images
AI

Why trust is a big question at the Elon Musk-OpenAI trial

Anthony Ha
Listen on Apple Podcasts Listen on Spotify

Lawyers for Elon Musk and OpenAI made their closing arguments this week, and now it’s up to jurors to decide whether OpenAI did anything wrong as it’s transformed into a slightly-more-for-profit organization. 

But as Kirsten Korosec, Sean O’Kane, and I noted on the latest episode of TechCrunch’s Equity podcast, a big theme in the trial’s final days was whether OpenAI CEO Sam Altman is trustworthy — for example, Musk’s attorney Steve Molo grilled Altman about whether statements he’d made during congressional testimony were truthful.

Kirsten noted that Musk has made plenty of misleading statements of his own, and that trust isn’t just an issue for Altman.

“This is a fundamental question [for] a lot of tech journalists, policymakers, and more and more consumers, about all the AI labs,” she said. “It’s really come down to trust, because we don’t have the insight, necessarily — these are all privately held companies, there’s a lot behind the veil still.”

Keep reading for a preview of our conversation, edited for length and clarity.

Anthony Ha: [The end of the trial] led to this really provocative headline from one of our writers, Tim Fernholz, [that] just says, “Who trusts Sam Altman?” Does anyone want to take a stab at answering this? 

Kirsten Korosec: Yeah, Anthony, I’m going to throw it right back to you. Do you trust Sam Altman? 

Amazon's Steve Schmidt on AI agents gone rogue (Live at HumanX) | Equity Podcast
0 seconds of 19 minutes, 31 secondsVolume 0%
 

Anthony: It's an interesting question because it feels like something that's kind of a wild question to discuss in a journalistic context, but actually that's the core of the trial, in a lot of ways. 

Sean O’Kane: That’s not a yes.

Anthony: And it actually seems to be [at the] core of understanding so much of what's happened at OpenAI, especially this big executive power struggle that they now call The Blip.

It just seems like a lot of people who've worked with Altman don't trust him. And he's acknowledged this a little bit, because he'll talk about the fact that he recognizes he's been conflict averse, telling people what they want to hear, and he's trying to work on that.

I mean, it sounds plausible, and I can understand how that can lead to misunderstandings in some situations. [But] I'm also a very conflict-averse person and I'd like to think that if any of this stuff went to trial, that people would not be asking, “Is Anthony Ha trustworthy?”

Sean: Still not a yes! 

Kirsten: I think that people would say that you are trustworthy. I will say that question, while provocative, doesn't just encapsulate what this trial was about. I would zoom out even more and say this is a fundamental question [for] a lot of tech journalists, policymakers, and more and more consumers, about all the AI labs. It's really come down to trust, because we don't have the insight, necessarily — these are all privately held companies, there's a lot behind the veil still.

Maybe when they all IPO, we can get a peek, but it is fundamentally about trust and misuse, and do we believe the intent? And what I would throw back is, sometimes the intent can be worthy, noble, and still misused. It can still end up as a bit of a shit show. I think it's more than who trusts Sam Altman — although that was very interesting in this trial — but more of that bigger question that we can apply to the entire industry. 

Sean: I'll say it: I don't trust him. But you know, I don't trust most people, so I guess that's just the baseline. 

We’ll see where this goes. The trial wraps up today. I've been very curious to hear how the jury decides this all. I think at the start of this, a big motivator of this was Elon Musk trying to sling mud, at a perceived rival and someone who he feels slighted him. And I don't know if we know enough yet to say that that was completely accomplished, and whether or not he has a shot at winning. But I think all these people came out of this looking a little bit worse. 

Anthony: And just to get specific, why this is coming up this week is that [Altman] was on the stand and he was basically getting grilled about some statements he's made in the past, in testimony to [Congress], basically saying he didn't have any equity in OpenAI. And that is not true because he had a stake through Y Combinator, which he used to run. And tried to brush that off by saying, “I assume that everybody understands what it means to be a passive investor in a VC fund.” And I think [Elon Musk’s] lawyer, somewhat fairly, said “Really? You think the congressman who was interviewing you knew that?”

Kirsten: Yeah, I mean, he was playing the whole semantics game. What I thought was so interesting about [this] is the style of how Sam Altman answered questions [compared to] Elon Musk on the stand. 

So Elon Musk, in many, many, many scenarios and many instances, we can point to the fact that he put something out on Twitter that was a lie or a bit of a fib, and on the stand corrected the record. So there's a history of, I would say, non-truthfulness-slash-lying, blatant or otherwise, in Elon Musk's world, but how he treated it was incredibly combative and very different than Altman who really took this [attitude of], “I'm working on it,” and tried to seem sort of affable and I don't know if it’ll work for him.

Because it really comes down to the core facts, and hopefully that's what the jury pays attention to. But I thought that that was really interesting — both being untruthful, but how they dealt with it was very different.

Loading the player...

Topics

, ,

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Anthony Ha

Anthony Ha is TechCrunch's weekend editor. Previously, he worked as a tech reporter at Adweek, a senior editor at VentureBeat, a local government reporter at the Hollister Free Lance, and vice president of content at a VC firm. He lives in New York City.

You can contact or verify outreach from Anthony by emailing anthony.ha@techcrunch.com.

View Bio
Keep reading
Cerebras Systems Andrew Feldman
Image Credits:Cerebras Systems
Startups

$60B AI chip darling Cerebras almost died early on, burning $8M a month

Julie Bort

Today, Cerebras Systems is a public company that sells AI chips for inference to giants like OpenAI and AWS. It held a blockbuster IPO on Thursday, with both of its co-founders billionaires, and ended the week worth about $60 billion.

But in 2019, when it was three years old, it came dangerously close to failure – incinerating a shocking amount of money. It was trying to solve a technical problem no one in the semiconductor industry thought could be done. 

“We were spending about $8 million a month,” founder CEO Andrew Feldman told TechCrunch of that period. “At this point, we had incinerated nearly $200 million trying to solve one technical problem.” 

Every few weeks, Feldman was forced to make the painful walk of shame to the board meeting to report another failure and more money burned. 

But he had no choice. Without a solution, Cerebras was dead anyway.

It was founded with an idea that was simple on paper. The microprocessor industry had spent its entire 50+ years making CPUs faster and cheaper by cramming more transistors onto a silicon wafer and dicing wafers into ever tinier pieces. But AI required so much compute power, many chips had to be strung together and then forced to communicate with each other. Cerebras’ founders believed turning a whole, even bigger wafer into one giant, powerful chip, would work faster. 

The problem was, no one had ever successfully done this before, for any reason, AI or not. Orchestrating that many microscopic electronic components onto a larger, but still thin, surface introduced compounding engineering problems. 

Everyone wants a piece of the enterprise AI pie | Equity Podcast
0 seconds of 31 minutes, 2 secondsVolume 0%
 

Once Cerebras crossed the first threshold of designing the mega chip and then manufacturing it with TSMC, the team hit the real roadblock. 

They couldn’t solve “packaging.” This involves everything after manufacturing the silicon itself: adhering it to a motherboard, getting power to it, dealing with heating and cooling as well as the pipes that would deliver and return data, Feldman said. 

Cerebras’ chips “were 58 times larger. We were using 40 times as much power as anybody had ever used,” he said. There were no premade heat sinks. No vendors. No manufacturing partners. The brightest minds in microprocessor engineering had tried for decades to build such big, yet more dense chips, and failed. 

The Cerebras team was left with trial and error in which “we destroyed an enormous number of chips” and an enormous amount of cash. But without functional packaging, the chip was useless. 

After exhaustive analysis of each failure, the team finally solved enough problems: how to cool it and move data around. In one instance, they had to invent their own machine that could bolt-in 40 screws simultaneously to secure the wafer to a board without cracking it. 

Feldman still remembers the day in July 2019 when it all, miraculously, worked.

They installed the packaged chip into a computer, turned it on and the entire founding team (pictured below) “just stood in the lab and stared at it,” he said. “Watching a computer run is about as exciting as watching paint dry. But there we were watching lights flashing on the computer, stunned that we'd solved this.” 

“That was one of the greatest moments of my life,” he said. That's significant, because this same founding team had previously built and sold a pioneering cloud server startup, SeaMicro, to AMD for $334 million in 2012.

Cerebras Systems founding team in 2015: Andrew Feldman, Gary Lauterbach, Michael James, Sean Lie and Jean-Philippe Fricker
Cerebras Systems founding team in 2015: Andrew Feldman, Gary Lauterbach, Michael James, Sean Lie and Jean-Philippe FrickerImage Credits:Cerebras Systems

The day the chip finally worked was also about two years after OpenAI had talked to Cerebras acquiring it, which Feldman confirmed to TechCrunch occurred like the publicly revealed emails said it did. 

Those talks fell through amidst growing squabbling among the OpenAI founders, several of whom are angel investors in Cerebras. 

Today OpenAI is a customer and a partner, having loaned Cerebras $1 billion secured by warrants. Those warrants conditionally grant OpenAI about 33 million shares of Cerebras’ stock, the S-1 discloses. (33 million shares are worth over $9 billion at Friday's closing price of $279.) 

Interestingly, Cerebras also agreed to not sell its wares to specific OpenAI competitors as part of that loan deal. Feldman wouldn’t confirm that the obvious company this involves: Anthropic. He did, however say that restriction is temporary. 

“It's limited in time, and it was designed to make sure that we could get OpenAI the capacity,” he said.

The truth was, Cerebras hasn’t yet grown big enough to handle multiple fast-growing model makers anyway.  He likened selling AI compute capacity to an all-you-can eat buffet. Instead of trying to stuff itself on all potential customers, “We're going to work with part of the buffet only, and we're going to get comfortable with that, before we attack the rest,” he said.

Topics

, , , ,

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Julie Bort
Julie Bort is the Startups/Venture Desk editor for TechCrunch.

You can contact or verify outreach from Julie by emailing julie.bort@techcrunch.com or via @Julie188 on X.
View Bio

Latest in Startups

See More
Loading the next article
Some areas of this page may shift around if you resize the browser window. Be sure to check heading and document order.