Helion Raises $465M Series G at $15.5B Valuation to Power Microsoft's Data Centers with Fusion

For decades, nuclear fusion was the technology that was always 20 years away. In 2026, for the first time in the history of the field, it is beginning to look like it might be three years away — and an investment round announced on June 4 is the clearest signal yet that major institutional capital is betting on that timeline being real.

Helion Energy, the Redmond, Washington-based fusion startup backed by OpenAI CEO Sam Altman, has raised $465 million in Series G funding at a $15.5 billion valuation, led by Thrive Capital. The round brings Helion’s total capital raised to $1.5 billion and puts the company on an accelerated path toward its most consequential commercial milestone: delivering fusion electricity to the grid for Microsoft by 2028.

The Microsoft Agreement and What Depends on It

The Microsoft deal, announced in May 2023 and structured as a power purchase agreement, committed Helion to delivering at least 50 megawatts of fusion-generated electricity to Microsoft within five years. If Helion misses the deadline, the company owes Microsoft financial penalties — a contractual structure that amounts to a forced public accountability mechanism for a fusion startup that would otherwise operate in scientific ambiguity.

That accountability has a new urgency in 2026. Microsoft, like every other hyperscaler, is facing acute pressure to source clean, reliable power for the AI data centers it is racing to build. The traditional grid cannot keep pace with AI’s power demands — a single large-scale GPU training cluster can consume as much electricity as a small city. As a result, nuclear power has experienced a renaissance in the tech sector: Microsoft has contracted with Constellation Energy to restart Three Mile Island’s Unit 1 reactor, and Google has invested in small modular reactor development. Against that backdrop, Helion’s promise of commercially viable fusion — an energy source with effectively unlimited fuel and no long-lived radioactive waste — occupies a unique position in the clean energy portfolio of major AI companies.

A Technology That Defies Convention

What distinguishes Helion from most fusion competitors is its approach to converting fusion energy into usable electricity. The dominant model in fusion research relies on capturing the extreme heat produced by plasma reactions to boil water into steam, which then drives conventional turbines — essentially the same process used in coal or fission nuclear plants. Helion takes a fundamentally different route.

The company’s design uses rapidly oscillating magnetic fields to compress and heat plasma in what it calls a field-reversed configuration. When fusion occurs, the expanding plasma pushes against those same magnetic fields, and the resulting induction allows electricity to be extracted directly from the magnets — a process the company compares to regenerative braking in electric vehicles. The approach, if it works as designed, eliminates the steam-turbine conversion step and offers the theoretical potential for higher overall efficiency.

Helion’s seventh-generation prototype, called Polaris, has achieved several industry firsts: it became the first privately funded fusion machine to operate with deuterium-tritium fuel — the most energetically favorable fuel combination — and earlier this year surpassed its own temperature record by exceeding 150 million degrees Celsius, roughly ten times the temperature at the core of the Sun.

The gap Helion has not yet publicly closed is net energy gain — demonstrating that the fusion reaction produces more energy than the machine consumes to initiate it. The company acknowledges this milestone remains ahead, but argues that the path from its current plasma temperature records to net energy gain is primarily an engineering challenge rather than a fundamental physics barrier.

Investor Roster and Competitive Context

The Series G drew a roster of sophisticated deep-tech investors. Thrive Capital led, joined by new investors including Alta Park Capital, Anti Fund, BoxGroup, Lux Capital, Peak XV Partners, and Bill Ford, the executive chairman of Ford Motor Company. Existing backers including Capricorn Technology Impact Funds, Lightspeed Venture Partners, Mithril Capital, Dustin Moskovitz’s Good Ventures Foundation, and SoftBank Vision Fund 2 also participated.

The valuation of $15.5 billion — more than triple the company’s previous round valuation — reflects both Helion’s technical progress and the changed macro environment for AI-adjacent energy plays. Data centers are now widely projected to represent one of the fastest-growing segments of electricity demand globally over the next decade, and any credible technology that can deliver reliable, carbon-free baseload power commands a premium.

Helion is not without competitors in the private fusion race. Focused Energy raised $240 million this year, Thea Energy closed a $100 million round, Inertia Energy raised $450 million in a Series A, and Type One Energy secured $250 million in a Series B. Commonwealth Fusion Systems, backed by Google and other major investors, is pursuing a different technical approach using high-temperature superconducting magnets. The breadth of the field suggests that investors believe multiple approaches may prove viable — or that no single approach has clearly won yet.

Scientific Skepticism and the Path Forward

Helion is not without critics in the scientific community. Some plasma physicists question whether the company’s field-reversed configuration can sustain the plasma conditions required for net energy gain long enough to be commercially useful. Skeptics also note that Helion publishes infrequently in peer-reviewed journals, limiting independent scientific scrutiny of its claims.

The company’s response, in essence, is that the proof will be in the plant. The Microsoft agreement — with its financial penalty structure for failure to deliver — functions as a form of external accountability that replaces the peer review process with a harder commercial deadline. If Helion is generating electricity for Microsoft data centers in 2028, the scientific debate will be rendered largely academic.

For the AI industry, the stakes extend beyond any one company. The accelerating power demands of frontier model training and inference have already strained regional power grids, prompted multi-billion-dollar nuclear restart deals, and pushed cloud providers to explore geothermal and offshore wind at unprecedented scale. Fusion, if it arrives on Helion’s timeline, would represent the most consequential energy technology development since the commercial nuclear era began in the 1950s — and the most direct answer yet to the question of whether AI’s insatiable appetite for electricity can be sustainably fed.