Fusion gain explained: why there is no single Q

Fusion gain is a ratio: energy out divided by energy in. It is usually written Q. The energy out is rarely in dispute. The energy in is where the arguments are, because a fusion experiment spends energy at several stages, and each stage gives a different answer.

A headline that says a device “produced more energy than it used” is true for some choice of denominator and false for others. This benchmark therefore never records a bare gain. Every gain figure is filed as one of six kinds, each with its numerator and denominator stated, and the gain page shows them side by side.

The six kinds of gain

Target gain (Q_target). Fusion energy released, divided by the laser energy delivered to the target. Used for laser-driven inertial fusion. The denominator counts the light that reached the target, not the electricity needed to make it. See target gain.

Capsule gain. Fusion energy released, divided by the energy the fuel capsule actually absorbed. In indirect-drive laser fusion, most of the laser energy heats the surrounding hohlraum and never reaches the capsule, so capsule gain is always larger than target gain. Quoting it as “the” gain overstates performance. The dataset holds no capsule gain figure. See capsule gain.

Plasma gain (Q_plasma, also called scientific gain or Q_sci). Fusion power divided by the external heating power put into the plasma. The magnetic-confinement counterpart of target gain. It leaves out the electricity used by the magnets, the heating systems’ own losses and everything else on site. See plasma gain.

Equivalent gain. The gain a deuterium-only (D-D) plasma would have reached had it been run on deuterium-tritium (D-T) fuel. It is a calculation, never a measurement, and the benchmark marks it as simulated and never holds it as achieved. See equivalent gain.

Engineering gain (Q_eng). Gross electricity generated divided by all the electricity the plant takes in, including the power it recirculates to run itself. This is the number a power station must beat. No fusion device has generated electricity, so no device has an engineering gain above zero. See engineering gain.

Wall-plug gain. Fusion energy released divided by the total energy the facility drew from the grid for the shot. It is almost never quoted by operators, and it is typically two to three orders of magnitude below target gain. See wall-plug gain.

Gain	Numerator	Denominator
Target	Fusion energy yield	Laser energy delivered to the target
Capsule	Fusion energy yield	Energy absorbed by the capsule
Plasma	Fusion power	External heating power coupled into the plasma
Equivalent	Modelled D-T fusion power	External heating power coupled into the plasma
Engineering	Gross electrical output	Total electrical input, recirculating power included
Wall-plug	Fusion energy yield	Total facility grid energy for the shot

One shot, two gains: NIF, 5 December 2022

The National Ignition Facility (NIF) at Lawrence Livermore National Laboratory fired shot N221204 on 5 December 2022. The peer-reviewed account in Physical Review Letters gives:

• 2.05 MJ of laser light delivered to the target
• 3.1 MJ of fusion energy released, with an uncertainty of ±0.16 MJ

Target gain: 3.1 ÷ 2.05 = 1.5. This was the first time any fusion experiment released more energy than was delivered to its target, and the benchmark records it as achieved. See the NIF facility page.

Wall-plug gain: LLNL states that NIF’s power conditioning system stores about 400 MJ of electrical energy for each shot. 3.1 ÷ 400 = 0.00775, which the dataset holds as 0.0078, rounded up because it is an upper bound.

That second figure is an upper bound, not an exact value. The 400 MJ is the energy held in the laser’s capacitor banks. It leaves out cooling, controls, cryogenics and making the target, all of which draw more power. The true wall-plug gain is lower still. For that reason the record stays at status “claimed” rather than “achieved”: it is a ratio built from two sources, and its denominator is an approximation.

The two figures describe the same shot on the same day and differ by a factor of about 200 (1.5 ÷ 0.0078 ≈ 192). Neither is wrong. They answer different questions. Target gain asks whether the fuel physics works. Wall-plug gain asks whether the facility as a whole came out ahead. Reporting one while implying the other is a common way fusion results are overstated.

LLNL calls the December 2022 shot “ignition”, a word with more than one meaning. LLNL’s ignition page defines it physically: the point where heating by the fusion reactions’ own alpha particles “overcomes the cooling effects of x-ray losses, electron conduction, and implosion expansion”. The US Department of Energy’s announcement of the result used a simpler test, “more energy from fusion than the laser energy used to drive it”, which is target gain above 1.

NIF has since gone further. The dataset holds a target gain of 4.13 on shot N250406, 7 April 2025: 8.6 MJ of fusion energy from 2.08 MJ of laser light. That figure is LLNL’s own and is held as claimed. LLNL also lists a shot on 20 June 2026 with 7.9 MJ of yield and a target gain of about 3.8; the dataset holds both as claimed.

Magnetic confinement: JET in 1997 and 2023

For tokamaks and other magnetic devices, the comparable figure is plasma gain.

JET, 31 October 1997. Shot 42976 on the Joint European Torus produced about 16 MW of fusion power. The Wurzel and Hsu compilation gives the heating power as about 26 MW, for a plasma gain of 0.63. The same shot is widely quoted as Q = 0.67, using 24 MW as the input. The fusion power is not in dispute; the difference is entirely in how the heating power is counted. The benchmark records 0.63 because that is the value whose denominator it can state. It remains the highest plasma gain in the dataset. See the JET facility page.

JET, 3 October 2023. Pulse 104522, in JET’s final deuterium-tritium campaign, released 69.26 MJ of fusion energy over five to six seconds (UKAEA says five, EUROfusion six) from about 0.2 milligrams of fuel. That is the largest fusion energy from a single magnetic-confinement pulse in the dataset. It is not a gain record. Wurzel and Hsu give its scientific gain as 0.37, lower than in 1997. The pulse was longer, not more efficient. The dataset records the energy but does not yet hold a gain figure for this pulse.

The contrast is the point. A record in fusion energy (a numerator) is not a record in gain (a ratio). JET’s 2021 and 2023 records were records in megajoules. On Wurzel and Hsu’s figures (0.33 for the 2021 pulse, 0.37 for 2023), neither came as close to breakeven as the 1997 shot.

For comparison, TFTR in the United States reached a plasma gain of 0.28 on shot 76778, 27 May 1994.

Equivalent gain: JT-60U

Japan’s JT-60U ran on deuterium, not D-T. Its operators calculated what the same plasma would have produced with D-T fuel. The dataset holds one such figure, 0.185, from about 2002, for a plasma in which the current was driven entirely without the central transformer. It is not JT-60U’s highest: the widely quoted equivalent gain of about 1.25, from a 1998 discharge, is not in the dataset because no primary source for it has been captured. The 0.185 figure is filed as equivalent gain with the simulated flag set, and it appears on no ranking of achieved results. See the JT-60U facility page.

Design targets

ITER is designed to produce 500 MW of fusion power from 50 MW of heating power, a plasma gain of 10. Commonwealth Fusion Systems’ SPARC has a stated mission of plasma gain above 2, and a published projection of about 11. Both are held as planned. A planned figure is permanently excluded from every achieved leaderboard: a target is not a result.

How to read a gain claim

1. Find the denominator. If the source does not say what was divided by, the figure cannot be compared with anything, and the benchmark holds it as claimed with a note.
2. Check the fuel. A D-D result converted to D-T is equivalent gain, a calculation.
3. Check whether it was measured. A design point or roadmap figure is planned, not achieved.
4. Separate energy from gain. Megajoules released is a numerator on its own.

The methodology page sets out how each status is assigned.

Sources

• Abu-Shawareb et al., “Achievement of Target Gain Larger than Unity in an Inertial Fusion Experiment”, Physical Review Letters 132, 065102 (2024)
• Lawrence Livermore National Laboratory, NIF power conditioning system
• Lawrence Livermore National Laboratory, Achieving fusion ignition
• US Department of Energy, DOE National Laboratory makes history by achieving fusion ignition (December 2022)
• Wurzel and Hsu, “Continuing progress toward fusion energy breakeven and gain as measured against the Lawson criteria”, arXiv:2505.03834 (2025)
• EUROfusion, JET’s final tritium experiments yield new fusion energy record
• UKAEA, JET’s final deuterium-tritium experiment results revealed one year on
• ITER Organization, Goals
• Wikipedia, JT-60 (equivalent gain of 1.25 in 1998)
• Creely et al., “Overview of the SPARC tokamak”, Journal of Plasma Physics 86 (2020)