Microsoft bets big on nuclear fusion by 2028

After embracing artificial intelligence, Microsoft is taking another gamble on a promise from OpenAI’s CEO for one more moonshot goal—nuclear fusion. As CNET reports, Microsoft announced it has entered into a power purchase agreement with a startup company called Helion Energy that is slated to go into effect in 2028. Unlike AI’s very immediate realities, however, experts suspectbelieve the project’s extremely short timeframe and technological constraints make this timeline unrealisticcould easily prove disastrous.

Nuclear fusion is considered by many to be the end-all be-all of clean, virtually limitless energy production. Compared to fission reactions within traditional nuclear power plants that split atoms apart, fusion occurs when atoms are forced together within extremely high temperatures to produce a new, smaller mass atom, thus generating comparatively massive amounts of energy in the process. Researchers accomplished important fusion advancements in recent years, but a sustainable, affordable reactor has yet to be designed. What’s more, many experts estimate achieving this milestone won’t happen without “a few decades of research,” if ever.

Helion was founded in 2013, and received a $375 million investment from OpenAI CEO Sam Altman in 2021, shortly after it became the first private company to build a reactor component capable of reaching 100 million degrees Celsius (180 million degrees Fahrenheit). The optimum temperature for fusion, however, is roughly double that temperature. Meanwhile, Altman’s OpenAI itself garnered a massive partnership with Microsoft earlier this year, and has since integrated its high-profile generative artificial intelligence programming into its products, albeit not without its own controversy.

[Related: Physicists want to create energy like stars do. These two ways are their best shot.]

Helion aims to have its first fusion generator online in 2028. This generator would theoretically provide at least 50 megawatts following a one-year ramp up period—enough energy to power roughly 40,000 homes near a yet-to-be-determined facility location in Washington state. From there, Microsoft plans to pay Helion for its electricity generation as part of its roadmap to match its entire energy consumption with zero-carbon energy purchases by the end of the decade. As CNBC notes, because it’s a power purchase agreement, Helion could face financial penalties for not delivering on its aggressive goal.

In 2015, Helion’s CEO David Kirtley estimated their company would achieve “scientific net energy gain” in nuclear fusion within three years. Within nuclear fusion research, this energy gain refers to the ability to viably emit more power than it takes to produce. When asked this week by MIT Technology Review if Helion met those goals, a representative declined to comment, citing competitiveness concerns, but said its “initial timeline projections” had assumed the company would raise funds faster than it ultimately managed.

“We still have a lot of work to do,” Helion CEO David Kirtley also admitted in a statement released Wednesday,  but we are confident in our ability to deliver the world’s first fusion power facility.”