Energy
Fusion: reading the milestones properly
Several genuine records have been set recently, and the gap between scientific breakeven and a power station is larger than the coverage suggests.
Fusion research has produced real milestones in the past few years. Interpreting them requires knowing which quantity is being compared to which.
The definitions that matter
The confusion in most reporting comes from a single term used for several different ratios.
Scientific gain. Fusion energy produced divided by energy delivered to the fuel.
This is the figure in the widely reported inertial confinement result, where fusion output exceeded the laser energy that reached the target.
It is a genuine scientific first and a real threshold.
Wall-plug gain. Fusion energy produced divided by electrical energy drawn from the grid to run the facility.
In the same experiment, the lasers required vastly more electrical input than they delivered to the target, because the laser system's efficiency is low.
On that measure the experiment consumed far more than it produced, by a very large factor.
Engineering gain. Net electricity delivered to the grid divided by electricity consumed.
This has never been achieved by any fusion experiment, and it is the only one that matters for a power station.
When a milestone is announced, identifying which of these is being reported resolves most of the argument.
The two main approaches
Magnetic confinement. Holding a hot plasma in a magnetic field for extended periods.
Tokamaks are the dominant design, with stellarators — more complex to build, potentially better at steady-state operation — as the main alternative.
The large international project under construction is the flagship of this approach, and its schedule has extended repeatedly.
Inertial confinement. Compressing a small fuel capsule rapidly with lasers so it fuses before it disassembles.
The facility that achieved scientific gain uses this approach, and its primary purpose relates to weapons physics rather than energy.
Repetition rate is the central problem for energy: a power plant would need several shots per second, and the facility fires a few times per day.
What remains unsolved for a power plant
Continuous operation. Experiments run for seconds to minutes. A plant runs for months.
Tritium supply. The most common fuel cycle requires tritium, which is radioactive with a short half-life and does not occur naturally in useful quantities.
The plan is to breed it within the reactor from lithium, using neutrons from the fusion reaction itself.
Breeding blankets have never been demonstrated at scale, and the required breeding ratio leaves little margin. This is arguably the largest unsolved engineering problem in the field and receives the least coverage.
Materials. The reactor wall is bombarded by high-energy neutrons that displace atoms and transmute the material, embrittling it and making it radioactive.
Materials that survive a plant lifetime under that flux do not yet exist and cannot be fully tested without a suitable neutron source, which is itself a facility that has not been built.
Heat extraction. Converting neutron energy to electricity at high efficiency, in a structure that also has to breed tritium and survive.
Cost. The unavoidable question. A fusion plant is a complex nuclear facility. Whether the electricity it produces could compete with alternatives is not established, and history with complex nuclear construction is not encouraging.
The private sector
Substantial private investment has entered the field, pursuing varied approaches — high-temperature superconducting magnets allowing smaller tokamaks, field-reversed configurations, magnetised target fusion and others.
The strongest argument for optimism is high-temperature superconductors, which permit much stronger magnetic fields. Confinement improves steeply with field strength, so a stronger magnet allows a much smaller and cheaper machine.
That is a genuine step change in the engineering rather than an incremental improvement, and magnet demonstrations have met their targets.
The counter-argument is that none of the remaining problems — tritium breeding, materials, continuous operation — is addressed by better magnets.
How to read a timeline
Fusion timelines have slipped consistently for seventy years, and the joke about it always being thirty years away is earned.
What has changed recently: scientific gain achieved, superconducting magnets demonstrated, private capital at scale, and several machines under construction.
What has not: any demonstration of tritium breeding, of materials surviving lifetime neutron flux, or of net electricity.
A reasonable position is that a demonstration plant producing net electricity in the 2030s is plausible for the best-funded efforts, that commercial deployment is a 2040s question at the earliest, and that the technology is therefore not a solution to near-term decarbonisation regardless of how it develops.
That is not an argument against funding it. It is an argument against counting on it.
