Helion

Helion has spent more than a decade on a contrarian bet: that fusion isn't a physics mystery waiting to be solved, but an engineering problem you grind down by building, testing, and iterating. Founded in 2013 in Washington State by David Kirtley and a small group of plasma physicists, it has now built seven prototypes, each one pushing a single system closer to a commercial power plant.

The wager is finally getting expensive to doubt. Helion has raised roughly $1.5 billion, its latest round valuing it at $15.5 billion, and it has signed the first commercial fusion power agreement in history with Microsoft. Its seventh machine, Polaris, ran deuterium-tritium fusion in 2026, and the company has broken ground in Malaga, WA on Orion, which it calls the world's first fusion power plant.

Most fusion designs make heat, boil water, and spin a turbine, losing energy at every step. Helion skips all of it. Its machines fire two rings of plasma at each other at a million miles per hour, compress the collision with pulsed magnets to over 100 million degrees Celsius, and then recover the energy directly as electricity when the expanding plasma pushes back against the magnetic field.

The company calls this a pulsed, non-ignition system, and it thinks of the physics like regenerative braking in an electric car. The payoff it is chasing is smaller, cheaper, factory-buildable machines that deliver always-on, carbon-free baseload power to the grid.

Helion frames fusion as the missing piece of the clean-energy grid: always-on power that doesn't care about weather, doesn't run a chain reaction, and doesn't leave long-lived radioactive waste. Its fuel is deuterium pulled from water plus helium-3 the company plans to produce inside its own machines, which sidesteps the scarcity and geopolitics that shadow fossil fuels.

The machine also shuts off the instant it stops being actively driven, and it only holds seconds of fuel at a time, which the company leans on as its core safety story.

Helion isn't selling to homeowners. Its customers are the buyers with enormous, around-the-clock electricity needs and a mandate to decarbonize: hyperscale data center operators and heavy manufacturers. That profile is exactly why its two named customers are Microsoft, chasing clean power for AI-era data centers, and Nucor, North America's largest steelmaker.

The surging demand for AI compute is the tailwind here, and Helion has said as much, pitching fusion as the answer to power buyers who want firm, carbon-free supply at scale.

Helion's product line is really a sequence of machines marching toward a power plant. Polaris, the seventh prototype, is a 19-meter device with a 50-plus megajoule capacitor bank and 15-plus tesla magnets, built to show it can recover electricity directly from fusion. In 2026 it became the first privately funded machine to run bulk deuterium-tritium fusion and hit plasma temperatures around 150 million degrees Celsius.

Orion is the next step and the real business: the commercial plant now under construction in Malaga, WA, designed to deliver electricity to the grid. Helion also runs HERCULES, a $4 million program funding 20 research institutions on the enabling tech that scaled fusion will need.

B2B power supply. Helion sells electricity to large industrial and hyperscale buyers through long-term power purchase and plant-development agreements, and is building the fusion power plants that fulfill them.

Helion is playing a different game than most of the well-funded fusion field. Where Commonwealth Fusion Systems, TAE, and the ITER-style tokamak world pursue steady-state confinement and heat-to-turbine conversion, Helion runs a pulsed, field-reversed-configuration machine and converts plasma energy straight to electricity.

That direct-recovery, helium-3 architecture is its main claim to a smaller, cheaper, faster-to-deploy plant. The catch, which its own critics press on, is that the design is unusual enough that few outsiders can independently check it.

Helion's case rests on hardware, not slideware. It has built and tested seven prototypes, each retiring a specific risk: Grande hit 4 tesla compression, Trenta reached plasma temperatures above 9 keV and compressed FRC plasmas past 8 tesla, and the company demonstrated direct magnetic energy recovery at over 95% round-trip efficiency across more than a million pulses.

The headline result came in 2026, when Polaris became the first privately developed fusion machine to demonstrate measurable deuterium-tritium fusion and reached roughly 150 million degrees Celsius. That milestone, more than any funding number, is what the company points to when asked whether the approach is real.

What supporters like is concrete: Helion actually builds machines, it has real milestones like the D-T fusion result and direct energy recovery, and it landed the first commercial fusion contract when it signed Microsoft. The engineering-first, ship-it culture reads as refreshingly grounded in a field full of promises.

The criticism is just as consistent. Helion has missed its own deadlines, including a loud 2021 promise to show net electricity by 2024, and it rarely publishes in peer-reviewed journals, so physicists can't independently vet the approach. Some plasma experts call the commercial timeline audacious to the point of science fiction, and co-founder John Slough, who later split with the company, has warned that its aggressive compression could drive plasma instabilities severe enough to lose confinement before fusion pays off.

Helion closed a $465 million Series G in June 2026 at a $15.5 billion post-money valuation, roughly triple its prior mark, bringing total money raised to about $1.5 billion. Thrive Capital led the round, which the company says will expand U.S. fusion manufacturing and speed deployment.

The backer list is a who's who of tech money: Sam Altman, who seeded the company early, Dustin Moskovitz through Good Ventures, Reid Hoffman, SoftBank Vision Fund 2, Mithril, Capricorn, Lightspeed, Peak XV, Lux Capital, and Ford executive chairman Bill Ford. The valuation is now the bar the machines have to clear.

Helion has done the hard part of turning a fringe fusion concept into funded, permitted, under-construction infrastructure with named customers waiting on the other end. The Microsoft deal calls for delivering power later this decade, and Orion is the machine that has to make good on it.

The risk is the same one skeptics keep naming: the timelines have slipped before, the physics of direct recovery and helium-3 fuel is unproven at commercial scale, and the company's reluctance to publish leaves outsiders trusting its own results. If Orion delivers electricity to the grid, Helion rewrites the energy map. If it slips again, a $15.5 billion valuation gets very heavy.

Helion is engineering-led and operator-built, and it hires for people who want their hands on real hardware: capacitors, magnets, pulsed-power systems, and plasma diagnostics. Its four stated values, build with urgency, take ownership, execute rigorously, and deliver hard truths, all circle back to the founding conviction that fusion gets solved by iterating faster than anyone thinks possible.

The work is concentrated in Everett, WA, with the Orion build site in Chelan County, and most roles are onsite and hardware-bound. It's a mission-first shop that openly tells candidates the work is really hard, which is rather the point.

Helion discloses salary ranges on many of its roles, and they skew high for a hardware company. Engineering bands run from roughly $108,000 to $320,000 a year, with operations and other functions landing in a similar spread and a handful of hourly technician roles alongside them. All figures are in U.S. dollars for Washington-based positions.

The bigger draw is ownership. Every Helion employee receives stock options plus a 5% 401(k) match and performance-based annual increases, so the upside is tied directly to whether the fusion bet pays off.

The registered entity is Helion Energy, Inc., operating out of Washington State, with a privacy policy and terms of use both effective May 31, 2021. Its stated safety posture is structural rather than procedural: fusion can't run a self-sustaining chain reaction, the machine stops the instant it isn't actively driven, and only seconds of fuel sit in the generator at any moment.

The company notes its machines produce manageable radiation while running, shielded with commercially available materials like concrete, and that fusion fuels such as tritium fall under existing dual-use export controls.

Helion's news cycle has turned from lab results to deployment. The through-line: a first-of-its-kind Microsoft power agreement in 2023, a record-setting deuterium-tritium result on Polaris, a construction permit and groundbreaking for Orion in Chelan County, and the $465 million Series G that reset its valuation.

Coverage runs from trade press like POWER Magazine and Nuclear Engineering International to Reuters, TechCrunch, and GeekWire, alongside the company's own technical blog.