In February 2026, Helion Energy disclosed that its Polaris prototype reactor had reached a plasma ion temperature of 150 million degrees Celsius — roughly ten times the temperature of the Sun’s core — and had become the first privately developed fusion machine to demonstrate measurable deuterium-tritium fusion reactions. Two months later the company confirmed continued operation at that temperature band, and earthwork has now begun on Orion, the commercial plant in Malaga, Washington, that has a signed contract to deliver at least 50 megawatts of electricity to Microsoft starting in 2028.

The headline number is real. The harder question — the question DOE physicists and hyperscaler procurement teams are asking simultaneously — is whether Polaris validates the unusual physics path Helion has chosen, and whether the Orion timeline survives contact with the rest of the engineering: pulsed-power switching, tritium handling, direct energy recovery at commercial scale, and Chelan County’s grid interconnect queue.

Why 150 Million Degrees Matters — And Why It Is Not the Finish Line

Plasma ion temperature is the easiest fusion metric to communicate and the hardest one to interpret in isolation. The Lawson criterion that governs net energy from a fusion plasma is a product: temperature, density, and confinement time. Helion’s pulsed Field Reversed Configuration architecture optimizes a different point on that surface than a tokamak or an inertial confinement target.

In a tokamak, plasma is held continuously at lower density for seconds at a time. In Helion’s approach, two FRC plasmoids are accelerated to roughly a million miles per hour from opposite ends of a linear chamber and collided in the center, compressing them adiabatically to fusion conditions for microseconds. The reactor pulses ten times per second. That cadence is closer to an internal combustion engine than to a power station. The 150 million °C figure is the peak ion temperature during the compressed pulse — the moment that matters.

The reason this design lets Helion think about 2028 grid delivery, when ITER will not see first plasma at full scale until 2034 and a tokamak-based commercial plant is a 2040s conversation, comes down to two physics choices and one engineering choice. The FRC operates at high beta, the ratio of plasma pressure to magnetic pressure, approaching unity. A tokamak runs at a beta of a few percent. High beta means Helion confines a hot, dense plasma using comparatively modest magnetic fields, which means smaller magnets, smaller buildings, and dramatically lower capital intensity per megawatt.

The Direct Energy Recovery Bet

The second choice is the one that confuses people most. Helion does not plan to make steam. The expanding fusion plasma pushes back against the magnetic field that contains it, and the changing flux induces a current in the surrounding coils — the same principle as regenerative braking in an electric vehicle, applied to a 150-million-degree plasmoid. Electricity comes out of the coils directly. There is no thermal cycle, no high-pressure steam loop, and no Carnot ceiling at ~33% conversion efficiency.

Direct energy conversion has been a theoretical attraction in fusion engineering for fifty years and a practical headache for almost as long. The pulsed-power electronics required to capture, condition, and dispatch megawatt-scale pulses ten times per second is non-trivial, and the IGBT-and-thyristor inventory required to do it cheaply did not exist outside fragmented industrial supply chains until the last decade. The convergence with EV-grade power electronics is one of the underappreciated reasons private fusion got fundable.

The D-T Surprise — And What It Says About the Fuel Cycle

Helion’s long-term fuel target is deuterium and helium-3. D-He3 is aneutronic — the dominant reaction produces a 14.7 MeV proton and a 3.6 MeV alpha particle, both charged, both directly capturable by the magnetic recovery system. No neutron flux means no activation of structural materials, no tritium breeding blanket, and no radioactive waste classification that triggers an NRC Part 50 licensing pathway. It is the cleanest fusion fuel cycle on paper.

Helion’s plan was always to generate He3 internally by fusing deuterium with deuterium in the same machine, capturing the He3 byproduct, and recycling it back into the D-He3 burn. The challenge is that D-D and D-He3 are both significantly harder to ignite than D-T. The D-T reaction has the highest cross-section of any practical fusion fuel and requires the lowest temperatures to initiate. It is the reaction every government program has chased since the 1950s for exactly that reason.

So the February announcement that Polaris had run D-T deserves a careful read. It is not a strategic pivot. It is a physics validation step. Demonstrating measurable D-T fusion at FRC compression confirms that the machine reaches genuine fusion temperatures and densities — not a synthetic neutron signal, not a back-of-the-envelope extrapolation, but a real reaction yield. It is the cleanest possible proof that the architecture works before scaling to the harder, cleaner, eventually aneutronic D-He3 cycle that defines the commercial product.

Orion: One Building, One PPA, One Grid Interconnect

The economic story is concentrated in a single sentence: Helion has a signed power purchase agreement, with Constellation Energy as the power marketer, to deliver at least 50 megawatts of fusion electricity to Microsoft from a single plant in Malaga, Washington, beginning in 2028. There is no second customer. There is no second plant. If Orion runs, Helion sets the commercial price floor for the entire private fusion industry. If it does not, the industry’s most exposed timeline slips and the procurement narrative resets.

Site work began in mid-2025. Chelan County granted the Conditional Use Permit for the fusion generator building in October 2025. Earthwork is underway. The Malaga site was chosen because Chelan County PUD has hydroelectric transmission already configured for industrial-scale offtake from the Columbia River system, which collapses interconnect risk that would otherwise dominate the schedule. Helion is not waiting for a new substation. It is plugging into one.

The AI Data Center Pull That Made the PPA Possible

Microsoft did not sign the Helion PPA in 2023 because fusion is fashionable. It signed because by 2028 it expects to be running an AI training and inference fleet that, on current trajectories, will require firm dispatchable carbon-free power at a scale the western interconnect cannot supply from any conventional source. The same logic produced Microsoft’s $16 billion, 20-year contract for the Three Mile Island Unit 1 restart, Google’s first-of-a-kind small-modular-reactor fleet PPA with Kairos Power, Amazon’s $20 billion Susquehanna data-center campus acquisition, and Meta’s 6.6 gigawatt nuclear request for proposal that closed earlier this year.

By the end of 2024, hyperscalers had contracted more than 16 GW of nuclear capacity, almost all of it tied to AI compute. The U.S. Department of Energy now expects U.S. data center electricity consumption to roughly double by 2028. Power certainty has displaced land, network latency, and even labor as the binding constraint on AI infrastructure siting. A 50 MW fusion plant in Washington is a rounding error against the gigawatt-scale data center campuses Microsoft is planning — and that is exactly the point. The PPA is not the volume play. It is the proof point that makes the next ten plants financeable.

The Competitive Landscape, Honestly

Commonwealth Fusion Systems has raised more money than Helion and is building SPARC in Devens, Massachusetts, targeting Q greater than 1 in 2027 and a commercial plant, ARC, in Virginia in the early 2030s. SPARC’s REBCO high-temperature superconducting magnet program is the most fundable thing in fusion outside of Helion’s PPA, and CFS deserves credit for de-risking magnet technology that the entire tokamak ecosystem now depends on.

TAE Technologies is pursuing a different FRC architecture aimed eventually at proton-boron fuel, which would be even cleaner than D-He3 but is significantly harder to ignite. General Fusion and Zap Energy are pursuing magnetized target fusion and sheared-flow Z-pinch respectively, both with credible physics and longer commercial timelines. ITER, the international tokamak in southern France, is the reference machine but is now scheduled for first plasma in 2034 and D-T operation in the 2040s. National Ignition Facility at Lawrence Livermore demonstrated scientific breakeven on inertial confinement in December 2022 — a historic milestone, not a power-plant pathway.

None of these competitors has signed a power purchase agreement for commercial electricity. Only Helion has staked its credibility on a delivery date that arrives before most utilities issue their next integrated resource plan. That is the bet.

What Could Still Break Between Polaris and Orion

Three risks dominate. The first is sustained Q. Polaris demonstrated fusion reactions; what it has not yet publicly demonstrated is engineering breakeven — more electricity recovered from the direct energy conversion system than was consumed to fire the pulse train. That number is the actual product, and it remains private. The second is repetition rate. A commercial plant at 50 MW requires sustained pulsing for thousands of hours between maintenance windows, which is a fatigue regime that no fusion machine on Earth has accumulated.

The third is the regulatory pathway. The Nuclear Regulatory Commission has signaled that fusion devices generally fall outside the Part 50 fission framework and into a lighter byproduct-materials regime under Part 30. Washington State has been cooperative. None of this has been tested at commercial scale in the United States. The first plant to push power onto a regional balancing authority’s grid is also the first test of fusion grid interconnection in this regulatory environment.

The Bottom Line for the Energy Industry

If Orion delivers 50 megawatts to Microsoft on schedule in 2028, the consequences fan out fast. Hyperscaler procurement teams will treat fusion the way they currently treat advanced fission — as a real option in capacity planning, not a research-budget line item. Long-duration energy storage economics shift. Natural gas peaker plants face their first credible challenger on dispatchability. The Genesis Mission and the DOE Milestone-Based Fusion Development Program get the validation data they need to commit a much larger second tranche.

If Orion slips by a year, the industry absorbs it. If it slips by three, the private fusion thesis takes serious damage, and the conventional view that fusion is always thirty years away reasserts itself. Either way, the next thirty-six months in Malaga will be the most consequential thirty-six months in fusion since the National Ignition Facility achieved ignition.

Polaris hit 150 million degrees in February. The relevant question is no longer whether Helion’s physics works. The relevant question is whether the engineering, the supply chain, the workforce, and the grid all converge on the same date. That date is 2028. The building is in Malaga. The world is watching.

---

Timothy Porritt is founder of Porritt Inc., building AI-powered tools for process safety, engineering compliance, and industrial operations. Based in Salt Lake City, Utah. Porritt Inc. consults with energy operators and AI infrastructure developers on PSM, refinery operations, and the convergence of advanced energy and industrial AI. To discuss a project, request a demo, or explore Genesis Mission teaming, reach out at porrittinc.com.