How far away is fusion power?

Why fusion matters

Fusion joins light atomic nuclei and releases energy in the process (LLNL). Its appeal is a possible large source of electricity, but that depends on plants producing more power than they use, at a reasonable cost. The experiments so far have not shown that. This article separates what has been measured from what is only claimed or hoped for.

How it works

The fuel is heated until it becomes plasma, a hot gas, and the experiments described below try to hold that plasma together long enough for fusion to take place.

Two main approaches are being tested. Tokamaks use strong magnetic fields to hold plasma in a doughnut-shaped ring (UKAEA). Laser facilities, such as the National Ignition Facility (NIF) in the US, instead fire lasers into a small cylinder that holds a fuel capsule. The lasers produce X-rays that make the capsule implode (LLNL).

Ignition is the point where heating from fusion in the hot centre of the fuel outweighs the processes that cool it, so the burn can sustain itself (LLNL). For NIF, LLNL also uses a simpler working test: producing more fusion energy than the laser energy delivered to the target (LLNL).

What has been demonstrated

These are measured results from experiments.

Ignition at NIF. A 2022 paper in Physical Review Letters reports a NIF experiment that reached ignition by nine different formulations of the Lawson criterion. In that experiment the target gain was 0.72, from 1.37 MJ of fusion energy for 1.92 MJ of laser energy, so it did not reach "scientific breakeven" (Abu-Shawareb et al., 2022). LLNL dates the achievement of ignition, in the sense of more fusion energy than laser energy on target, to 5 December 2022 (LLNL).

Target gain above one. On 5 December 2022, NIF produced 3.15 MJ of fusion energy from 2.05 MJ of laser energy delivered to the target, according to LLNL (LLNL). The peer-reviewed paper on that shot gives 3.1 MJ from 2.05 MJ, a target gain of 1.5, and calls it the first laboratory demonstration of exceeding scientific breakeven (Abu-Shawareb et al., 2024).

Repeated shots. LLNL reports that NIF achieved ignition for an 11th time on 20 June 2026 (LLNL). Where LLNL states a target gain for these shots, the figures are 2.44 on 23 February 2025, 4.13 on 7 April 2025, 1.74 on 1 October 2025 and about 3.8 on 20 June 2026. Other entries give only yield and laser energy, and some are lower; for example, on 8 October 2023, 1.9 MJ of laser energy gave 2.4 MJ. On 7 April 2025 one shot gave 8.6 MJ, plus or minus 0.45 MJ, from 2.08 MJ on target. On 20 June 2026 a shot gave 7.9 MJ, plus or minus 0.4 MJ.

Magnetic confinement at JET. The JET tokamak in the UK released 69 MJ of fusion energy in a single pulse of about five seconds, using 0.2 milligrams of fuel (UKAEA). The pulse, lasting 5.2 seconds, was run on 3 October 2023 and announced on 8 February 2024. It beat JET's 2021 record of 59 MJ, and the energy was released in the form of fast neutrons (Max Planck Institute for Plasma Physics). JET was a research facility, not a power station, and it ended operations at the end of 2023 (UKAEA).

What is only reported

These are design figures, official counts or investment totals. They are not measured outputs of a working plant.

ARC design. A paper published online on 4 June 2026 in the Journal of Plasma Physics sets out the physics basis for the ARC Version 3A design from Commonwealth Fusion Systems (CFS). It says the design is meant to produce about 1.13 GW of fusion power and deliver at least 400 MW of net electrical power to the grid (Hillesheim et al., 2026). Across a range of physics models and input assumptions, the paper reports predicted fusion power of roughly 600 to 1300 MW. These are design targets and model predictions, not measurements.

Device count and investment. The IAEA's World Fusion Outlook 2025, published in October 2025, counts more than 160 fusion devices that are operational, under construction or planned (World Nuclear News). In his foreword, as relayed by World Nuclear News, the IAEA Director General says global private investment has surpassed USD 10 billion (World Nuclear News). We have relied on World Nuclear News's account rather than the IAEA report itself.

Company timelines. The ARC paper states that CFS plans to build ARC as the first fusion power plant at a site in Chesterfield County, Virginia, by the early 2030s (Hillesheim et al., 2026). This is a company target, as stated in a paper published in June 2026, not a demonstrated result.

What is speculative

Grid electricity by a given date. The US Department of Energy's finalised fusion roadmap, announced on 9 June 2026, is described as a national strategy to support fusion pilot plants and commercial fusion power in the mid-2030s (DOE). This is a government goal, not a commitment. DOE says its ability to support the milestones and timelines depends on future public-private partnerships and congressional appropriations, and that the roadmap does not commit DOE to specific funding levels.

Company plant dates. The CFS aim of an ARC plant by the early 2030s, noted above, is also a target that has not yet been tested by construction or operation.

Cost per kilowatt-hour. The sources we reviewed give no verified cost figure for fusion electricity, so no figure is given here.

Limits and open problems

• Energy gain against energy input. NIF's target gain compares fusion yield with laser energy on the target. It does not count the electricity used to run the laser or the facility, so it is not net electricity. None of the sources we reviewed shows a fusion plant supplying net power to a grid.
• Repeatability. The NIF results are individual shots spread over about three and a half years. The JET record was one pulse. The sources do not show a rapid, repeatable sequence of shots suitable for a power plant.
• Schedule. ITER's revised baseline, announced in July 2024, has research operations starting in 2034 and deuterium-tritium operation in 2039. The Max Planck Institute for Plasma Physics describes the 2039 date as a four-year delay for deuterium-tritium operation (IPP).
• Engineering and cost. The ARC paper says engineering analysis, and analysis of economic impacts, are outside its scope (Hillesheim et al., 2026).
• Tritium and materials. JET's final campaign used deuterium and tritium (UKAEA). The sources we reviewed do not show a closed tritium fuel cycle, or materials that last a plant's lifetime.

What to watch next

• ITER's progress against its 2034 and 2039 dates (IPP).
• New entries in LLNL's list of ignition shots, and whether shots become more frequent as well as more energetic (LLNL).
• Whether DOE's mid-2030s goal is matched by funding (DOE).
• Publication of detailed ARC engineering and economic analysis, which the physics overview leaves outside its scope, and progress against the CFS early-2030s target (Hillesheim et al., 2026).