OpenAI in Talks to Buy 5 Gigawatts of Fusion Electricity by 2030

Sam Altman stepped down from the board of Helion Energy on Monday after more than a decade as an investor and director, saying his dual roles had become untenable as OpenAI and Helion begin exploring a significant partnership. He will keep his financial stake in the company and recuse himself from deal negotiations between the two.

"Sam has played an integral role in Helion's development," Helion CEO David Kirtley said in a statement to TechCrunch. "This decision enables Helion and OpenAI to partner on future opportunities to bring zero-carbon, safe electricity to the world. We look forward to continuing to work with him in this new capacity."

Altman has been backing Helion since 2015. His departure from the board is the cleanest governance move available when the chairman of a supplier's board becomes the CEO of a potential customer — especially at the scale now reportedly being discussed.

Axios reported Monday, citing a single source familiar with the situation, that the talks would give OpenAI 12.5% of Helion's total electricity output: five gigawatts by 2030 and 50 gigawatts by 2035. Helion flatly did not confirm this. A company spokesperson told Reuters that "beyond the previously announced deals with Microsoft and Nucor, Helion has not made any new customer announcements." OpenAI did not respond to Reuters' request for comment on the Axios report.

That means the headline deal numbers — 5 GW and 50 GW — remain a single-source report. They are worth examining on their face. Helion has said each of its commercial reactors will generate 50 megawatts. OpenAI's 12.5% share would imply Helion needing to field roughly 800 reactors by 2030 to deliver 5 GW to OpenAI, and around 8,000 reactors by 2035 to deliver 50 GW. Helion's first commercial plant isn't built yet. Its Polaris prototype, currently operating in Everett, Washington, hit 150 million degrees Celsius in February — progress, but still short of the roughly 200 million degrees Celsius the company believes commercial operations require.

That gap between the Axios figures and Helion's current development stage is the honest story here. If the numbers are real, they imply a rate of fusion reactor deployment with no precedent in the history of energy infrastructure. If they're aspirational — numbers put on paper to anchor a long-term relationship — then what's actually happening is that OpenAI is securing optionality on fusion power that may or may not materialize.

The strategic logic isn't hard to follow. Microsoft, Google, and Amazon have all spent the past few years locking in nuclear and clean energy deals as data center demand strains the grid. OpenAI has been the conspicuous holdout, and Helion is a natural partner given Altman's decade-long investment. Microsoft signed its own Helion power purchase agreement in 2023, targeting delivery starting in 2028.

This is the second time Altman has run this play. In April 2025, he stepped down as board chair of Oklo, a small modular reactor startup that had merged with his acquisition company AltC. At the time, Oklo's COO Caroline Cochran said the move was intended to allow Oklo to explore strategic partnerships with leading AI companies, "including potentially with OpenAI," according to CNBC. The pattern is deliberate: Altman uses his capital and board presence to identify energy companies early, then exits the governance structure when OpenAI is ready to become a customer.

Helion was founded in 2013 by Kirtley alongside John Slough, Chris Pihl, and George Votroubek. Unlike most fusion startups, which capture heat from fusion reactions and run it through steam turbines, Helion's design converts fusion energy directly into electricity using magnets — a more efficient path if it works. The company raised $425 million in a Series F in January 2025, valuing it at $5.4 billion.

The board departure is confirmed and on the record. The power deal is Axios-reported and denied by Helion. Both things can be true simultaneously: Altman leaving the board is a prerequisite for a deal, not proof that a deal exists.