Nuclear Fusion Timeline 2026: The Race to Put Fusion Power on the Grid

What does fusion “Q” actually mean? Reports often say a machine “achieved net energy” without specifying which energy boundary. NIF’s December 2022 shot delivered 2.05 megajoules of laser energy to its target and got 3.15 megajoules of fusion energy back — a real, independently verified result. But the lasers themselves drew roughly a hundred times that much electricity from the wall to fire, because today’s laser systems are highly inefficient at converting grid power into light. So the same experiment was simultaneously a landmark for target-level scientific gain and nowhere close to facility-level net energy. Whenever this page uses “gain,” it specifies which boundary is meant.

CFS, spun out of MIT, builds compact high-field tokamaks using high-temperature superconducting (HTS) magnets. SPARC is the demonstration machine, meant to prove the physics case for net energy gain in a compact device; CFS installed SPARC’s first superconducting magnet in early 2026, and is now targeting first plasma around 2027 after earlier, more optimistic dates slipped. ARC is the separate, proposed commercial plant — do not confuse the two. CFS describes ARC as designed around approximately 400 MW of net electricity, enough for on the order of 150,000–300,000 homes by the company’s own estimate, targeting a Virginia site with Dominion Energy as a strategic partner. Under the DOE’s Milestone-Based Fusion Development Program, CFS has cleared preconceptual-design and technology-roadmap milestones for ARC, backed by five peer-reviewed papers examining the plant’s physics. CFS has raised over $4 billion to date, including an $863 million round in 2026 that brought in Nvidia’s venture arm alongside existing backers Google, Eni, Breakthrough Energy Ventures and a consortium of Japanese industrial firms; Google and Eni have separately signed power-purchase agreements covering more than half of ARC’s planned output — before the plant is built. ARC is being designed to deliver that output; it does not yet exist.

Helion’s approach is not a tokamak. It uses a pulsed field-reversed configuration (FRC): magnetically confined plasmoids are accelerated and collided, compressed by pulsed magnetic fields, and the fusion reaction’s energy is captured largely through direct electromagnetic conversion rather than heating a coolant to drive a turbine. Polaris is the current, seventh-generation prototype; in a company announcement made February 13, 2026, Helion reported that Polaris achieved measurable deuterium-tritium fusion in January 2026, with plasma temperatures around 150 million°C — about three-quarters of what Helion believes it will need for commercial operation, by its own account. Orion is Helion’s proposed first grid-supplying plant, under construction in Malaga, Washington. It carries a 2023 power-purchase agreement with Microsoft for at least 50 MW starting in 2028, ramping to full output over roughly a year. In June 2026, Helion became the first fusion company ever to hold a Radioactive Materials License and Radioactive Air Emissions License from Washington State — a real regulatory milestone, but one that confirms safety-programme readiness, not that net power has been proven or that Microsoft is yet receiving electricity.

ITER, in Cadarache, France, is a joint research project of the EU, US, Russia, China, Japan, South Korea and India, built to demonstrate the scientific and technological feasibility of fusion — not to generate electricity. ITER will never sell power to a grid. Its mission is proving that a burning, self-sustaining plasma can be controlled at scale, informing every power-plant design that follows it. On July 28–29, 2026, ITER lowered the sixth of nine roughly 1,100-tonne vacuum-vessel sector modules into the tokamak pit, putting two-thirds of the core in place about six months ahead of schedule; the final sector is now expected around mid-2027. Under ITER’s 2024 baseline, Start of Research Operation begins in 2034, deuterium-deuterium plasma operation in 2035, full magnetic energy in 2036, and operation with the actual fusion fuel — deuterium-tritium — from 2039. That is roughly four years later than the 2016 baseline, attributed to the pandemic, quality issues, and, ITER’s own leadership has acknowledged, overly optimistic early planning for a first-of-a-kind machine.

EAST (Experimental Advanced Superconducting Tokamak), in Hefei, is China’s long-running research device, known for repeated long-duration, high-confinement plasma records that inform reactor design worldwide — it is a research tool, not a power plant. BEST (Burning Plasma Experimental Superconducting Tokamak), under construction at the same Hefei complex, is China’s next step: a machine explicitly designed to demonstrate actual burning deuterium-tritium plasma. In August 2026, China’s Institute of Plasma Physics completed what it describes as the world’s largest fusion magnet — a 582-tonne superconducting coil for BEST — alongside a 400-tonne-plus Dewar base, the largest vacuum component China’s fusion programme has produced. China is targeting BEST construction completion by the end of 2027, with a net fusion-power gain and electricity-generation demonstration targeted around 2030 — state media has called this “lighting humanity’s first nuclear-fusion-powered lamp.” That framing is a government target, not a confirmed outcome.

Run by the UK Atomic Energy Authority, STEP (Spherical Tokamak for Energy Production) aims to build a prototype fusion power plant at West Burton, Nottinghamshire, the site of a former coal power station. In 2026 the programme moved from concept design into delivery: a public consultation ran January 14 to March 11, a construction partner was appointed for a £200 million site redevelopment, and up to 10,000 jobs are projected by 2030 tied to the redevelopment. Plant assembly and infrastructure construction are planned to begin in the 2030s once planning permissions are secured, with first operations targeted in 2040, demonstrating at least 100 MW of net energy “as soon as practicable.” STEP is currently the only major government programme with an explicit 100 MW-plus grid target and a public date, deliberately set later than several private-company targets.

Tokamak Energy builds compact spherical tokamaks using high-temperature superconducting magnets. Through 2026, its ST40 device is undergoing a $52 million upgrade — jointly funded with the US DOE and the UK’s Department for Energy Security and Net Zero — adding lithium plasma-facing systems and RF heating. Its Demo4 magnet set reached 11.8 tesla at −243°C carrying seven million ampere-turns, a real HTS-engineering result. The company’s power-plant plan, ST-E1, targets up to 200 MWe delivered to a grid in the early 2030s — a company target pursued through the US Milestone-Based Fusion Development Program, alongside a magnet-technology role in STEP. A separate demonstration device, ST80-HTS, had not been confirmed complete as of mid-2026 and no longer features prominently in the company’s public roadmap.

The Sun fuses hydrogen using immense gravitational pressure at its core, at “only” around 15 million°C. Earth has no equivalent gravity well, so terrestrial reactors substitute much higher temperatures — often above 100 million°C — confined by magnetic fields or inertial compression, and must build every part of the physical environment gravity provides for free inside a star.

A viable power plant must also pay for its reactor, magnets, fuel cycle, turbines or conversion hardware, cooling, ongoing maintenance, component replacement after neutron damage, financing, grid connection and staff. No commercial fusion fleet exists anywhere, so real market electricity costs remain unknown; where companies publish cost targets, they are projections, not observed prices, and this page does not repeat any of them as fact.

Is fusion safe? Fusion avoids the risk of a runaway chain reaction because the reaction stops as soon as confinement is lost — a real, structural safety advantage over fission. But it is not risk-free: it involves tritium handling, neutron-activated materials, powerful magnetic fields, cryogenic systems and standard industrial hazards. Does fusion create radioactive waste? Yes, though the profile differs from fission — neutron activation of structural materials produces waste that is generally shorter-lived, but “zero waste” is not an accurate claim for any current design.

India has been a full ITER member since the project’s 2006 founding agreement, and its domestic fusion research is coordinated by the Institute for Plasma Research (IPR) in Gandhinagar, home to India’s own SST-1 superconducting tokamak. Through ITER-India, India delivered the project’s single largest component: the 3,850-tonne cryostat, a 30-metre vacuum chamber housing the entire tokamak, designed by ITER-India and manufactured with Larsen & Toubro. India also manufactured the in-wall shielding — roughly 8,900 individual blocks providing neutron shielding between the vacuum vessel’s double walls — through Avasarala Technologies of Bengaluru, plus cooling-water systems, high-voltage power supplies and diagnostic neutral-beam systems. India does not currently have a domestic commercial fusion power-plant programme; its role to date is as a major engineering and manufacturing contributor to the international research effort.

The old joke persists because fusion schedules have genuinely slipped for decades — ITER’s own baseline has moved roughly four years later than its 2016 plan, and CFS’s SPARC date has moved from earlier public estimates to around 2027. First-of-a-kind hardware, novel materials, complex regulation and global supply chains all slow first attempts; that is normal engineering history, not unique dishonesty. What has genuinely changed since the “30 years away” era: high-temperature superconducting magnets that shrink reactor size and cost, far greater computing power for plasma simulation and control, private capital exceeding $14 billion cumulatively, and, in 2026, the first-ever regulatory approval of a fusion plant. What has not disappeared: the fuel cycle, materials survivability, heat exhaust and economics. Both things are true at once — real progress, and continued reasons for scrutiny of any specific date.

AI did not start fusion research, which predates it by seven decades. But GPU-cluster data centres now demand enormous, continuous (“firm”) electricity, straining grids in ways utilities are actively planning around — and that demand is one factor behind Microsoft’s Helion contract and Google/Eni’s ARC power-purchase agreements. No data centre runs on fusion electricity today; these are forward contracts for plants still under construction. Separately, AI tools are genuinely used inside fusion labs — for real-time plasma-instability prediction and control, and for materials and reactor-design simulation — but AI has not “solved fusion,” and no credible lab claims it has.

Fusion has already crossed milestones that once seemed remote: high-performance plasmas, a genuine laboratory ignition result, advanced superconducting magnets, and now a private machine reporting deuterium-tritium fusion and the first-ever regulatory licenses for a fusion plant. But a commercial power plant demands far more than a good plasma shot: durable materials, a working fuel cycle, functioning heat exhaust, financing, regulation, and — the only test that actually matters commercially — electricity leaving the site often enough and reliably enough for a grid to depend on it.

The defining fusion breakthrough will not be another record plasma temperature. It will be a power plant that keeps sending electricity to the grid, day after day. That milestone has not happened yet, anywhere, as of August 2026.

US Department of Energy — Fusion Science & Technology Roadmap (June 2026) · ITER Organization — Two-Thirds Tokamak Core Assembly · Commonwealth Fusion Systems — SPARC · Helion Energy — Polaris D-T Fusion Milestone · Helion Energy — Orion Regulatory Licenses · Tokamak Energy · UK STEP Fusion Programme · Fusion Industry Association — 2026 Global Fusion Industry Report · ITER-India — India’s Contribution to ITER