On May 14, 2026, a 48-ton steel ring was lowered by crane into the SPARC tokamak pit at the Commonwealth Fusion Systems campus in Devens, Massachusetts. It was the second half of the vacuum vessel that will eventually hold deuterium-tritium plasma above 100 million degrees Celsius. With that lift, SPARC reached 75% structural completion. The external cryostat closure and fuel-line integration remain — and the company is still publicly committed to first plasma in 2026 and net energy gain (Q>1) in 2027.
Three months earlier, in February 2026, Helion Energy’s Polaris prototype outside Seattle did something quieter but arguably more consequential. On its first shot with tritium fuel, Polaris produced a plasma that crossed 150 million degrees Celsius and generated measurable deuterium-tritium fusion. Helion became the first private company in the world to legally possess tritium for a fusion demonstration and the first to burn it.
Two privately funded fusion machines. Two milestones inside one quarter. Neither was a press-release \”breakthrough\” — both were the predictable next step in their respective build schedules. That is exactly what should make engineers and physicists pay attention. The fusion industry has spent a half-century delivering surprises; this quarter, it delivered scheduled work.
SPARC at 75%: What the Vessel Means Physically
SPARC is a high-field tokamak designed to produce roughly 140 MW of fusion power in pulses lasting around 10 seconds, with a target gain (Q) between 2 and 10. The architecture rests on one specific bet: rare-earth barium copper oxide (REBCO) high-temperature superconducting tape, wound into D-shaped toroidal field coils that produce a magnetic field of 12.2 tesla on axis. That field strength is what lets SPARC fit Q>1 physics inside a machine roughly 1/40th the volume of ITER.
The vacuum vessel matters because it is the boundary condition for every other system: the plasma-facing first wall, the cryostat, the divertor, the diagnostic ports, and the fuel handling lines all key off the vessel’s geometry. With both halves now in place, SPARC moves from civil construction into systems integration. That is a fundamentally different risk profile. Civil work risks are about heavy lifts, weld qualifications, and concrete schedules. Systems integration risks are about vacuum leaks, magnet quench protection, plasma diagnostics calibration, and tritium handling — all of which CFS has been derisking on smaller test articles for five years.
ITER, by comparison, is now targeting first plasma in 2039 and a deuterium-tritium burn no earlier than the mid-2040s. The gap between a 75%-complete private machine in 2026 and a public consortium targeting first plasma 13 years from now is the gap that the fusion industry has been waiting to demonstrate. It is now demonstrated.
Polaris in Tritium: Why Helion’s Shot Mattered More Than the Temperature Number
The headline from Helion’s February announcement was the temperature: 150 million °C, three-quarters of the way to what the company believes a commercial machine will need. The more important number is harder to put in a press release: Polaris is the first privately built machine to demonstrate the scaling laws extending from pure-deuterium operation into the D-T regime.
Helion does not use a tokamak. Polaris is a field-reversed configuration (FRC) machine that compresses two counter-streaming plasma rings into a single, denser plasma at the center of a linear chamber. Instead of holding plasma at ignition conditions for seconds, Helion pulses for microseconds and harvests electricity directly from the induced magnetic field changes — no steam cycle, no heat exchanger. That direct-conversion bet is unique in commercial fusion. It also means Helion’s path to grid power does not depend on tritium breeding, because the company’s commercial fuel cycle targets D-He3 produced from D-D side reactions.
The first D-T shot at 150 million °C is, in that sense, a proof point on the easier-to-confine fuel before the harder commercial fuel comes online. It tells physicists that compression efficiency, plasma stability, and direct conversion behavior in Polaris scale the way the models predicted. It is the kind of data point that derisks every subsequent shot.
Why the Hyperscaler Offtake Stack Now Matters
Commercial fusion needs an offtake market that will sign a 20-year power purchase agreement for a first-of-a-kind plant. Five years ago, that customer did not exist. Today it does, and it is the same customer for every credible fusion developer: hyperscale data center operators trying to build out AI capacity.
The procurement numbers tell the story. As of mid-2026, U.S. hyperscalers have committed to over 9.8 GW of nuclear capacity across 13 announced projects — Microsoft’s restart of Three Mile Island Unit 1 (835 MW for ~$16B over 20 years), Amazon’s 1.92 GW Susquehanna PPA with Talen Energy through 2042, Meta’s stack of TerraPower Natrium, Oklo Aurora, Vistra, and Constellation contracts totaling up to 6.6 GW, and Google’s earlier multi-SMR commitment with Kairos. U.S. data center load is projected to hit roughly 76 GW by year-end 2026, up from ~50 GW in 2024.
Fission gets first-mover advantage because the regulatory pathway exists. But every hyperscaler procurement team now has a fusion line item on their long-term planning deck. Helion’s existing agreement to deliver electricity to a Microsoft data center east of Seattle in 2028 is the template. CFS’s ARC plant — the commercial follow-on to SPARC — has begun siting work in Virginia, the same Northern Virginia / Mid-Atlantic corridor where the densest AI training clusters are being built.
This is the dual-use angle that matters for the broader energy industry: the same customer base willing to spend $650 billion on AI infrastructure capex in 2026 is the customer base that will absorb early fusion at any reasonable price, because their alternative is curtailment. That changes the financing equation for first-of-a-kind plants in a way that traditional utility offtake never could.
The Competitive Landscape, Stated Fairly
SPARC’s risk is integration time. The machine is at 75%; the path from structural completion to first plasma still requires the cryostat closure, magnet cooldown to 20 K, vacuum bake-out, diagnostic commissioning, and a careful plasma ramp campaign. Any of those steps can slip by quarters. The 2027 Q>1 target assumes a relatively smooth commissioning. The historical base rate for first-plasma slips in tokamaks is not zero.
Helion’s risk is repetition rate and pulse energy. A microsecond pulse that produces measurable D-T fusion is a milestone; a 10-Hz machine producing enough net electrical output to back a PPA is a different engineering problem. Polaris is the seventh-generation prototype, and the eighth-generation commercial machine is already in early design. The 2028 grid-delivery commitment to Microsoft is the most aggressive timeline in the industry. Helion will either prove the FRC direct-conversion architecture against that timeline or it will not.
ITER remains the most thoroughly instrumented and internationally validated fusion physics experiment ever attempted. Its 2039 first-plasma slip is not a verdict on fusion — it is a verdict on building a 23,000-ton tokamak via a 35-nation procurement consortium. The lesson is the same one the launch industry learned in the 2010s: privately financed, vertically integrated engineering teams beat distributed public consortia on schedule, not on physics.
What Net Energy in 2027 Would Actually Mean
If SPARC achieves Q>1 in 2027 — even at the low end of its 2-to-10 target — three things change inside 18 months. First, every fusion developer’s cost-of-capital drops, because the existence of a credibly commercial machine validates the asset class. Second, the hyperscaler PPA market begins pricing fusion as a real 2030s option rather than a 2040s option, which pulls capex schedules forward by half a decade. Third, the regulatory work on fusion-specific NRC pathways accelerates, because the agency now has a concrete machine to write rules against rather than a hypothetical.
None of that requires fusion to deliver gigawatts in 2027. It requires SPARC to produce a single pulse where the fusion power out exceeds the auxiliary heating power in. That is a far smaller engineering ask than the industry has been chasing for 50 years, and it now sits behind a cryostat closure, a magnet cooldown, and a plasma commissioning campaign — work that is fully scoped, fully funded, and underway.
The Physics That Matters
Fusion’s underlying physics has not changed in May 2026. What changed is the boundary between physics and engineering. SPARC’s REBCO-based 12.2 T field, Helion’s pulsed FRC compression, and Polaris’s D-T regulatory approval are all engineering achievements built on physics that has been understood for 40 years. The fact that two privately financed companies got those engineering pieces working inside the same quarter is the data point.
For executives evaluating long-horizon power supply, the operational frame is this: the probability that at least one commercial fusion plant is supplying grid electricity before 2032 has moved from \”very low\” to \”non-trivial\” in the last 12 months. That is not enough to base a capital plan on. It is enough to require a serious answer when the question comes up in board meetings — and the question is now coming up in board meetings.
The May 2026 milestones do not prove fusion. They prove the fusion industry can execute on schedule. For an industry that has spent five decades being late, that is the more important demonstration.
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. tracks the energy sector for dual-use applications across DOE Genesis Mission, ARPA-E, and SBIR/STTR funding pathways. Contact us to discuss fusion integration scenarios, hyperscaler offtake structuring, or PSM compliance for advanced energy assets.