Fusion startups are racing to put power on the grid before 2030
Fusion energy has spent seventy years as the technology that's permanently a decade away. In 2026, the honest version of that joke needs an update: several private companies have now poured real concrete, wound real magnets, and signed real power purchase agreements — but the underlying pattern hasn't changed. No commercial fusion plant is producing electricity for the grid, no company is within twelve months of doing so, and the original aggressive timelines these same companies announced have slipped, sometimes by years.
Both things are true at once, and separating them is the actual story: which 2026 milestones represent genuine engineering progress worth tracking, and which are marketing events dressed up as milestones.
Commonwealth Fusion Systems: the tokamak with a real construction schedule
Commonwealth Fusion Systems, the MIT spinout building the SPARC tokamak in Devens, Massachusetts, has the most conventional physics approach of the major players — a scaled-up version of the magnetically-confined plasma design used in decades of government fusion research, made viable at smaller size by high-temperature superconducting magnets. CFS completed SPARC's magnet ring by summer 2026 and is targeting first plasma in 2027, which would be the moment the company either demonstrates net energy gain from a compact tokamak or doesn't. Beyond SPARC, CFS says it will begin construction later this decade on ARC, a planned commercial power plant designed to produce 400 megawatts — but ARC's timeline depends entirely on what SPARC's plasma results actually show in 2027, not on the current construction schedule.
Helion Energy: the most aggressive bet, and the one already revised once
Helion is pursuing a different physics approach — pulsed, non-tokamak fusion aimed at direct electricity generation rather than heat-to-steam-turbine conversion — and has made the boldest public commitment of any fusion company: a 2023 power purchase agreement with Microsoft promising fusion electricity by 2028. Helion broke ground on Orion, described as the world's first commercial fusion power facility, in Malaga, Washington in July 2025. Orion reaching 2028 would make Helion the first private fusion company to deliver grid electricity, full stop — but a signed PPA is a commercial commitment, not a technical guarantee, and 2028 already represents one of several timeline adjustments the company has made since its earliest public statements.
TAE Technologies: fusion goes public via a SPAC
TAE Technologies took a different kind of step in 2026: a merger with Trump Media valued above $6 billion, making TAE the first publicly traded pure-play fusion company. That's a capital markets milestone, not a physics one — it gives TAE a public listing and retail investor access to a technology that, like its competitors, hasn't produced a watt of grid electricity yet. Google has separately backed both Commonwealth Fusion and TAE directly, part of a broader pattern of hyperscalers making direct bets on fusion suppliers rather than waiting for the technology to mature through traditional utility procurement.
Why the money keeps flowing despite the slippage
The fusion sector has raised billions from investors who are, on paper, betting against fusion's own seventy-year track record of missed deadlines. What's changed is the buyer. AI data center power demand has made hyperscalers — Google, Microsoft, and by extension Amazon and Meta — desperate enough for new firm, carbon-free generation capacity that they're willing to sign power purchase agreements years before a plant exists, effectively pre-financing construction the way airlines pre-order planes that haven't finished flight testing. That's a rational bet for a company like Microsoft with the balance sheet to absorb a failed PPA; it's a different risk profile for retail investors buying into TAE's public listing expecting fusion economics on a normal equity timeline.
What to actually watch instead of press releases
The signal worth tracking in 2027 is CFS's first plasma result from SPARC — a real physics milestone that will either validate the compact high-field tokamak approach or force a public reckoning with the design. Treat every PPA announcement as a financing and marketing event, not a delivery date; Helion's 2028 target and CFS's "later this decade" ARC timeline are both conditional on milestones that haven't happened yet, and both have already moved once from earlier public statements. And distinguish carefully between "net energy gain in a lab experiment" — which several fusion approaches, private and government, have already achieved in some form — and "continuously generating electricity a utility can dispatch to the grid," which no company, public or private, has done. Fusion in 2026 is closer than it was in 2016. It is not close.