Helion’s Polaris Fusion Machine: Electricity From Fusion in 2026?

For most of a century, fusion power has been the energy source that is always twenty years away. In 2026 that slogan still applies to most of the field. Helion Energy, based in Everett, Washington, is trying to make the timeline concrete: a working seventh-generation prototype called Polaris that has already hit two private-industry firsts, a commercial plant named Orion under construction in Chelan County for a Microsoft power deal, and a public hedge that full 50-megawatt operation will not land in 2028.

On February 13, 2026, Helion announced that Polaris had become the first privately developed fusion machine to demonstrate measurable deuterium-tritium (D-T) fusion and to reach plasma temperatures of 150 million degrees Celsius. Those results sit on top of a design that aims to recover electricity directly from expanding plasma — no steam turbine required — and on a 2023 power-purchase agreement that still commits Orion to grid connection and initial operations by the end of 2028, even as executives now say full power comes later.

Here is what Polaris is, what those February milestones actually mean, how Helion’s path differs from the big tokamak route, where the Microsoft deal stands in late 2026, and what still has to go right before electrons from fusion show up on a bill.

Polaris: the seventh and final prototype

Helion describes Polaris as its seventh and final prototype — a roughly 19-meter machine at the company’s Everett campus, north of Seattle. Public design specs list fuels of deuterium-deuterium (D-D), deuterium-tritium (D-T), and deuterium-helium-3 (D-He-3); a capacitor bank of more than 50 megajoules; peak magnetic fields above 15 tesla; inductive energy recovery; and thousands of diagnostics. The company began operating Polaris at the end of 2024.

The architecture is pulsed magneto-inertial fusion built around field-reversed configuration (FRC) plasmas. Two plasma rings form, accelerate, and merge; magnets then compress the combined plasma hard enough for fusion. As the magnetic field changes during plasma expansion, current is induced in the coils and can be captured directly as electricity on the capacitor bank. Helion emphasizes that this inductive recovery path does not require a steam cycle — a deliberate contrast with approaches that heat a blanket and drive turbines.

That framing matters for how Helion talks about goals. In 2021 the company used "net electricity" language for Polaris. Its current Polaris page prefers a more concrete target: demonstrate electricity from fusion by showing that fusion energy was made and that a portion was converted back to electricity on the capacitor bank. Polaris is not sold as a power plant. It is sold as the machine that de-risks Orion.

February 2026: D-T fusion and 150 million °C

Helion’s February 13, 2026 newsroom release put two claims at the center. First, Polaris demonstrated measurable thermonuclear D-T fusion — the company says it was the first privately developed fusion energy machine to do so, and that Helion was the first company to receive regulatory approval to possess and use tritium for demonstrating fusion energy production. Second, Polaris reached plasma temperatures of 150 million degrees Celsius, breaking Helion’s own private-industry record of 100 million °C set on the sixth-generation Trenta prototype.

Within the fusion community, about 100 million °C is often treated as a threshold for commercially relevant plasma temperature. Helion notes that D-T is a step on its testing program, not its commercial end state. Commercial systems are planned around deuterium-helium-3; D-T runs at lower temperatures for a given input energy and produces higher-energy neutrons useful for validating diagnostics, radiation safety and models before the company pushes toward D-He-3 conditions.

Director of Radiation Safety and Nuclear Science Michael Hua described the first D-T campaign in a company blog the same day: tritium injection modules installed under radiation-safety protocols, routine D-D pulses to check machine health, then a first D-T pulse that produced roughly a tenfold increase in neutron signal over comparable D-D pulses, followed by additional D-T pulses. Independent lines of evidence — organic scintillators, diamond detector arrays tuned to the 14.1 MeV D-T neutron signature versus 2.45 MeV D-D neutrons, and activation foils — were cited as confirmation. Outside reviewers quoted in the newsroom release, including researchers with Sandia and University of Michigan experience and longtime FRC expert Alan Hoffman, said they had reviewed diagnostic data consistent with D-T fusion and temperatures above about 13 keV.

Polaris is the machine that is supposed to turn fusion pulses into recovered electricity — and Orion is the plant that has to put some of that electricity on the grid.

Trenta, Vela and Tiny Merge: how Helion iterates

Polaris did not appear from nowhere. Trenta, Helion’s sixth prototype, was retired after a final campaign in 2023. Helion had reported Trenta plasmas above the 100 million °C commercial-threshold language, extended vacuum operations lasting on the order of sixteen months, and evidence of bulk deuterium-helium-3 fusion. Those results underwrote fundraising and the shift of effort to Polaris.

Alongside the headline machines, Helion runs subscale testbeds. Vela was a pulsed-power platform that, in company accounts, operated at 1 hertz and moved more than 11 gigajoules of total energy through a simulated Polaris compression magnet — a way to prove switching and magnet-drive hardware without tying every test to the full fusion device. Polaris Junior Formation ran from 2023 to 2025 to refine plasma formation in parallel with the main build.

In 2026 Helion has been integrating Tiny Merge, a thermonuclear fusion testbed GeekWire reported as roughly one-eighth the size of Polaris — on the order of eight feet long — aimed at faster iteration on plasma formation, merging dynamics and next-generation magnets. CEO David Kirtley has framed the subscale strategy as an engineering pattern closer to aerospace than to single-machine science campaigns: fail small, integrate what works, keep Polaris and Orion on schedule. That is the theory. The practice is that Orion’s calendar is already sliding on full power even as construction advances.

Orion, Microsoft and the 2028 hedge

In May 2023 Helion and Microsoft announced what both called the world’s first fusion power-purchase agreement: at least 50 megawatts of electricity from a Helion plant, with a 2028 start originally marketed as the year electricity would begin flowing to Microsoft. Orion, sited in Malaga in Chelan County, Washington, is the plant meant to meet that commitment. Helion’s Orion page now says the facility is designed to deliver fusion electricity to the grid, with initial operations beginning in 2028. The company has reported environmental review milestones, a conditional use permit, site construction beginning in 2025, and generator-building work advancing in 2026.

By September 2026, Axios and The Fusion Report were documenting a softer public timeline. Helion had removed website language that Orion would begin producing electricity by 2028 and replaced it with "begin initial operations in 2028," without defining that phrase. At the FusionX conference in Boston, public affairs director Jackie Siebens said Helion would "not, in no universe, have 50 megawatts hit full operation in 2028," and that full power would not come before 2029 or 2030. She also said the Microsoft contract requires Orion at least to be connected to the grid and begin operations by the end of 2028. Whether that 2028 connection includes supplying electricity to Microsoft remains unclear; Microsoft has declined to comment in that coverage.

The hedge is not a cancellation. Ground has been broken. Capital has continued to arrive — industry reporting around mid-2026 put a large new round on top of more than a billion dollars raised lifetime — and Helion still markets Orion as the first commercial fusion plant. But the gap between "initial operations / grid connection in 2028" and "50 MWe at full power in 2029–30" is now explicit. For a field that sells timelines hard, that honesty is useful context for readers tracking 2026 headlines about electricity from fusion.

How Helion’s path differs from tokamaks (briefly)

Most public fusion imagery still shows doughnut-shaped tokamaks: continuous magnetic bottles such as JET in the United Kingdom, EAST in China, or the under-construction SPARC device from Commonwealth Fusion Systems (CFS). Tokamaks aim for sustained high-performance plasmas, often with superconducting magnets and a thermal cycle that turns neutron heat into steam and turbines. CFS has raised far more capital than Helion and is building SPARC as a physics and engineering demonstration on a path toward a later ARC power plant — a different bet on time, scale and physics maturity.

Helion’s pitch is compactness, pulsed operation, fuel flexibility toward D-He-3, and direct inductive recovery. That path avoids some tokamak infrastructure and invents a different set of hard problems: high-repetition pulsed power, electrode and first-wall lifetime under repeated pulses, reliable fuel handling including tritium during the learning phase, and proving that recovered electricity is not a lab curiosity but a plant-relevant fraction. Neither path has put commercial fusion megawatts on a customer meter. A short contrast is enough: tokamaks are the mainstream magnetic-confinement story; Helion is a private FRC magneto-inertial story racing a PPA calendar.

Approach	Machine style	Electricity path (stated)
Helion (Polaris → Orion)	Pulsed FRC, ~19 m prototype	Direct inductive recovery; no steam cycle
Tokamak (e.g. JET / SPARC class)	Toroidal magnetic confinement	Usually heat blanket → steam / turbine
Helion commercial fuel goal	D-He-3 (with D-D / D-T in testing)	Same pulsed recovery architecture
Microsoft Orion target	Chelan County, WA plant	≥50 MW after ramp-up; 2028 initial ops / grid connection; full power later

Caveats: physics, manufacturing and calendar risk

Even with D-T neutrons and 150 million °C plasmas on the scoreboard, Polaris still has to finish the electricity-from-fusion demonstration Helion now emphasizes, then push temperatures and fuel mixes toward commercial D-He-3. Diagnostics can confirm fusion without proving a power plant. Materials, pulse rate, uptime and cost per megawatt are the boring items that decide whether Orion looks like a first-of-a-kind triumph or an expensive science project with a transmission interconnect.

Competition is real. CFS and other tokamak and stellarator efforts are not standing still; laser inertial programs and alternative magneto-inertial startups crowd the same venture narrative. Helion’s Microsoft deal creates both a customer signal and contractual risk — GeekWire has noted financial penalties tied to missing the 2028 connection-and-operations bar with Microsoft and partner Constellation. Fundraising that looked strong before a timeline hedge can look harder after one.

None of that makes the February 2026 milestones unimportant. Private D-T operation plus a new temperature record on a machine built for electricity recovery is a concrete step. It is also not the same thing as Microsoft data centers running on fusion power in 2028.

Where Helion sits next to other cleantech on aglena.com

On this site, the near-term grid story already includes Form Energy’s iron-air batteries aiming at 100-hour storage, sodium-ion batteries leaving the lab for grid niches, solid-state EV batteries still waiting for cars you can buy, perovskite tandems racing toward commercial roofs and fields, Iceland’s giant air filter finally hitting design pace, and the Tesla Roadster reveal as a separate mobility headline. Fusion is a different lane: firm, carbon-free generation rather than storage or vehicles — closer in spirit to first power from hot dry rock at Fervo’s Cape Station and to the reminder that permitted is not built for TVA’s first U.S. SMR than to a phone battery. Helion’s 2026 status belongs with the generation cluster — ambitious, capital-intensive and still one successful plant away from rewriting the "twenty years away" joke.

If Polaris converts fusion energy to banked electricity on schedule and Orion connects in 2028 while ramping toward 50 megawatts later, Helion will have moved the private fusion conversation from slides to a meter. If the electricity demonstration slips or Orion’s "initial operations" stay symbolic for years, the February temperature and neutron headlines will still matter for physics — and will look less like a commercial countdown. The useful 2026 scorecard is specific: D-T done, 150 million °C claimed, Everett still iterating with Tiny Merge, Malaga under construction, and 50 MWe pushed into 2029–30.

Helion Polaris FAQ

What is Helion’s Polaris machine?

Polaris is Helion’s seventh and final prototype in Everett, Washington — about 19 meters long, designed to demonstrate electricity from fusion via direct inductive recovery and to operate across D-D, D-T and D-He-3 fuels.

What did Helion announce in February 2026?

That Polaris demonstrated measurable deuterium-tritium fusion and reached plasma temperatures of 150 million degrees Celsius, which Helion describes as private-industry firsts.

What is Orion?

Orion is Helion’s first commercial plant, under construction in Malaga, Chelan County, Washington, intended to supply at least 50 megawatts under a 2023 power-purchase agreement with Microsoft.

Is fusion electricity coming in 2028?

Helion’s public language now emphasizes initial operations and grid connection by the end of 2028 under the Microsoft contract. Company officials have said full 50-megawatt operation will not happen in 2028 and is not expected before 2029 or 2030.

How is Helion different from a tokamak company?

Helion uses pulsed field-reversed configuration plasmas and aims to recover electricity inductively without a steam cycle. Tokamak projects typically pursue sustained toroidal plasmas and often couple fusion heat to conventional thermal power conversion.

Further reading