ADVERTISEMENT

Technology

Vinod Khosla Says ‘Fusion Will Be Real’ Within the Next Five Years

Vinod Khosla, founder of Khosla Ventures, during the Bloomberg Technology Summit in San Francisco. (David Paul Morris/Bloomberg)

(Bloomberg) -- A holy grail of clean power generation, nuclear fusion, will be a reality within five years, according to OpenAI backer and tech billionaire Vinod Khosla. 

A major booster of artificial intelligence who wrote the first venture capital check for Sam Altman’s OpenAI, Khosla said fusion will be key to meeting data center energy demand. 

Generating energy by smashing atoms together has long been a dream for investors and entrepreneurs eager for a source of limitless, clean baseload power. The technology saw a breakthrough in 2022 when scientists were able to get more energy out than energy put in for a fusion reaction.

But it’s struggled to reach viability, prompting skepticism it’ll become a reality. Soaring electricity demand due in part to the AI data center boom, though, has spawned a renewed interest in both moonshot approaches and traditional nuclear technology known as fission, which involves splitting large atoms instead. 

Major tech companies have poured money into fission recently; Microsoft Corp. is helping resurrect the shuttered Three Mile Island nuclear facility to power data centers, and competitors Alphabet Inc. and Amazon.com Inc. have invested in small modular reactors. (The latter rely on fission but have yet to be used outside a few small reactors.) But for Khosla, traditional nuclear isn’t as strong a long-term solution.

“I do think five, six years from now, we won’t be talking about fission because fusion will be real,” Khosla said in an interview for an upcoming episode of Bloomberg Green’s Zero podcast. He also said NIMBYism would also limit fission’s widespread resurgence. 

Both Amazon and Khosla see their divergent nuclear solutions reaching commercial scale by 2040. Amazon — along with billionaire financier Ken Griffin — have backed a small modular reactor project forecast to deploy 5 gigawatts by 2039. Khosla, meanwhile, projects that the world will see its first fusion plant by the early 2030s. He also sees coal plants as an ideal place to site fusion reactors.

“You’re not changing permitting, you’re not going through new grid connections, all this other stuff that takes a lot of time. I think it becomes a very short process relative to building new power plants,” he said.

Khosla has been a climate investor for over a decade, but he’s also an AI evangelist. Unlike many climate advocates who are concerned about the environmental impacts of AI, he thinks it will help boost demand for clean energy. 

“I’m really glad the demand side is solved now,” he said. “Power is a much harder problem because it needs longer-term planning.” He also thinks AI algorithms will become more efficient over time, requiring less energy.

Khosla said he has not invested in Altman’s efforts to raise money for a global AI infrastructure project, which includes an energy component. The OpenAI chief executive officer has spent part of the year working to build a global coalition of investors to fund AI industry infrastructure like data centers, expanding energy capacity and transmission. 

Altman has also invested separately in nuclear startups, including Helion Energy, which plans to open the world’s first fusion power plant by 2028. Khosla’s firm led an investment in startup Realta Fusion and backed Commonwealth Fusion Systems, considered among the leaders in the fusion race.

©2024 Bloomberg L.P.