When Sam Altman came to Helion Energy’s small Redmond, Wash., office in early 2014, carrying nuclear-fusion textbooks, the firm was primarily focused on research and development. By the time he departed, some days later, he had persuaded the fusion-energy firm to take a more aggressive approach to deployment, CEO David Kirtley says. A year later, Altman, who had co-founded OpenAI around the same time, invested $9.5 million in Helion and assumed the post of chairman. He invested an additional $375 million in Helion in 2021, making it one of the most significant personal bets in his multibillion-dollar portfolio.
Nuclear fusion was formerly a government-led effort, but it is now a private-capital race, with most of the funding coming from the same individuals who are developing energy-hungry AI and working toward the aim of producing systems with human-like intelligence, known as artificial general intelligence (AGI). According to a study from the European Union’s Fusion for Energy, total financing for the fusion-energy business has increased from $1.7 billion in 2020 to $15 billion by September 2025. Along with Altman, who has stated that the future of AI rests on an energy breakthrough, Helion’s investors include OpenAI sponsor SoftBank and Facebook co-founder and early Anthropic supporter Dustin Moskovitz. Nvidia has funded Helion’s competitor Commonwealth Fusion Systems (CFS). Google has also invested in another player, TAE Technologies. According to Troy Carter, head of Oak Ridge National Laboratory’s fusion-energy branch, artificial intelligence (AI) is a significant driver due to the energy requirements for data centers.
Recent advancements in engineering and financial interest have led to corporations offering grid electricity in the next years, rather than decades. They still need to show the technology works, but if fusion succeeds, it would give carbon-free electricity without the seasonal swings of solar and wind, or the long-lived radioactive waste of nuclear fission—a breakthrough that will not only cut power prices, but also change the possibilities.
Fusion, the same reaction that powers the sun, produces energy in the opposite way that modern nuclear power plants do, by uniting light atoms rather than dividing heavy ones. This happens deep into the core of a star in plasma, a super-hot, electrically charged gas. Recreating it on Earth proved to be the mother of all engineering challenges.
For decades, even when scientists were able to initiate a fusion reaction, it produced less energy than was necessary to heat the plasma, a phenomenon known as scientific breakeven. However, in 2022, researchers at Lawrence Livermore National Laboratory made history. They demonstrated for the first time a fusion reaction that produced more energy than was required to heat the plasma by temporarily crushing a small fuel pellet with huge lasers. No private enterprise has reached this milestone.
If and when they do, getting power onto the grid will need going one step further: producing not only more energy than was required to heat the plasma, but enough to run the entire generator, a process known as engineering break-even. Helion has the most optimistic timeline of any company aiming to reach that milestone. The business anticipates a commercial version of its machine to generate power by 2028 from a facility near Malaga, Washington, where building began in July. Helion has already agreed to sell 50 megawatts of fusion power to Microsoft and faces financial penalties if it deviates from the deadline.
Unlike other fusion attempts, which boil water to drive a turbine, Helion intends to generate power by hurling two plasma rings together at almost a million miles per hour. The impact would cause fusion, which disrupts a magnetic field and generates electricity. According to Kirtley, the arrangement now recaptures around 96% of its energy intake, similar to how an EV employs regenerative braking to recharge briefly throughout a drive. This puts it only a whisker away from breaking even. Polaris, Helion’s seventh-generation prototype, was expected to achieve engineering break-even in 2024. The prototype was turned on for the first time late that year. Kirtley refuses to release the results.
Kirtley, who credits Altman for driving him to “go faster and on a larger scale,” aspires to be the first to develop a fusion plant. “Our objective is to create one generator every day and deploy fusion systems worldwide. “And do it quickly.”
According to Carter of Oak Ridge, such bold thinking has legitimately moved fusion closer to reality. In 2020, he led a Department of Energy assessment that predicted a prototype nuclear-fusion reactor would be completed by the early 2040s, but he now believes the target may be met by the mid-2030s. Beyond wealth, AI is a valuable instrument for scientific advancement. “The advent of AI has made some very challenging problems in the plasma space more accessible,” says Nuno Loureiro, head of the Plasma Science and Fusion Center at MIT.
Even if Helion’s strategy falls behind schedule by a few years, it has the potential to achieve several world firsts. If this method fails, others will follow suit.
Pacific Fusion, a California-based firm, claims to have constructed a machine that would achieve engineering break-even using the same technique as the Lawrence Livermore National Laboratory’s device. It went public in 2024, with $900 million from investors including former Google CEO Eric Schmidt and Microsoft AI CEO Mustafa Suleyman.
CFS, which split out of MIT in 2018 to become the best-funded fusion company, is taking a different approach: designing a magnetic container to hold an ultra-hot plasma in place. CFS is developing a prototype that it says will reach scientific breakeven in 2027. The business is so confident that it has began construction on a commercial facility that will provide electricity to the grid in the early 2030s.
Google has already promised to purchase 200 megawatts. “[Having] these big hyperscalers behind us is really helpful,” says Brandon Sorbom, CFS’s co-founder and chief science officer, adding that it indicates to suppliers that offer superconducting magnets and other difficult-to-manufacture materials that “this isn’t a one-off science experiment.” Marc Benioff, co-chair and owner of TIME, is among Commonwealth Fusion Systems’ investors.
Meanwhile, New Zealand-based startup OpenStar began producing plasma in late 2024 with a modest $10 million in capital and has since received an additional $14 million. Its prototype reverses the “magnetic bottle” principle, placing an ultra-strong magnet at the reactor’s core and confining the plasma around it.
Carter is hopeful about the amount of firms striving to become industry leaders, but cautions that a high-profile failure may scare investors and undermine the field’s reputation. “You hope the hype doesn’t get too much; [that] a failure of one of the more visible companies does not pull the plug on progress that we have elsewhere,” he replies.
Fusion, however, cannot happen quickly enough for companies like Google and Microsoft. Both are developing new data centers to support AI, even though Microsoft aims to be carbon-negative by 2030 and Google aims for net-zero. AI data centers operate around the clock; without a breakthrough in energy storage, fluctuating wind and solar will be unable to meet that demand. And supply is scarce: US power output has not changed since 2010, when it was surpassed by China as the world’s greatest electricity producer. Even fossil fuels may struggle to scale as computing power increases. Altman and Nvidia’s Jensen Huang now refer to energy as the major bottleneck.
However, harnessing fusion power will have far-reaching consequences beyond simply powering data centers. It might be erected where energy is required rather than where the wind or sun conditions are optimal. With plentiful energy, global economic and geopolitical dynamics might be flipped upside down. “Most of our wars are fought over energy,” Carter argues. “If that’s no longer the driver, that changes things dramatically.”
