What commercial fusion means

Fusion power means fusing light nuclei to release energy and converting that energy to electricity at a price and scale someone will pay for. Those are three separate achievements, and the field has completed the first, is attempting the second, and has not begun the third. Keeping them apart is the main analytical discipline this topic requires, because the word “breakthrough” gets applied to all three interchangeably.

Where the science stands

Established Ignition — more fusion energy out of the target than the laser energy delivered to it — was achieved at the National Ignition Facility in December 2022 and has since been repeated. That settled the physics question. It did not produce net energy at the wall plug, because the lasers consume far more than they deliver to the target, and NIF is a weapons-physics facility rather than a power-plant prototype.

Frontier The commercial field is now substantial and diverse. Seventeen private fusion companies have each raised more than $100 million. The best-capitalised, a 2018 MIT spinout with roughly $3 billion raised, is building a compact tokamak whose central bet is high-temperature superconducting magnets: stronger fields allow a burning-plasma device far smaller than ITER. It was reported around 75% complete in mid-2026, targeting burning plasma in late 2026 or early 2027 and a gain factor above 2. Others pursue pulsed magnetic compression, sheared-flow Z-pinch, field-reversed configurations and laser inertial confinement — genuine technical diversity rather than one approach with many logos.

Established No commercial fusion plant has delivered electricity to a grid, and none is within a year of doing so. The leading company's commercial reactor is projected for the early 2030s at the earliest, with a utility power purchase agreement already signed against that output. Another has contracted to supply a technology company by 2028. These are commitments, not deliveries.

Frontier The field's own record on schedules is the most predictive datum available: essentially every major fusion milestone to date has taken longer than its roadmap projected, and the current generation's original targets have already slipped. Speculative Whether fusion becomes economically competitive with the alternatives available in the 2040s is a separate question from whether it works, and it is the one that decides whether any of this matters for carbon.

The binding bottleneck

Frontier The constraint is no longer plasma physics but the balance of plant: tritium breeding and handling, first-wall materials that survive neutron flux for years, heat extraction, magnet manufacture at volume, and the capital cost of all of it. Established Tritium in particular is a closed-loop problem — a deuterium-tritium plant must breed more tritium than it burns, and no facility has yet demonstrated that. Frontier Behind these sits the economic bottleneck: a fusion plant must compete against solar, wind and storage whose costs continue to fall.

Ethics and governance

Established Fusion carries genuine advantages that are often overstated but real: no long-lived high-level waste on the fission scale, no meltdown pathway, no weapons-usable fissile output. Activated structural materials are a real waste stream and a manageable one. Frontier The governance question is regulatory novelty — several jurisdictions have moved to regulate fusion separately from fission, which is defensible on hazard grounds and untested in practice. Frontier There is also an opportunity-cost argument worth taking seriously: capital and political attention directed at fusion are not available for deployment of technologies that could cut emissions this decade.

Timelines

  • 10 yr: Frontier burning plasma demonstrated; first pilot plants under construction; possible first grid electricity late in the window.
  • 25 yr: Frontier a small number of commercial plants plausible if tritium breeding and materials hold up.
  • 50 yr: Speculative material contribution to global generation, conditional on cost competitiveness that cannot yet be assessed.
  • 100 / 250+ yr: Speculative fusion as a civilisational baseload source, and the precondition for much of Categories I and II of this map.