← OrreryThe Ecliptic is an Orrery publication

Cover Story · The Reality Gap desk

Gain Is Only Half the Product

Laser fusion's record shot returned 4.13 times the light that reached its target. The laser behind it turns about 0.7% of its electricity into that light. A power plant needs the product of the two, and every driver efficiency above 0.7% in the public record is a design value, not a measurement.

~0.7%
wall-plug efficiency of the National Ignition Facility's laser: about 300 MJ drawn from the grid for about 2 MJ on target, the only full-scale measurement in the public record
4.13
NIF's record target gain (8.6 MJ out for 2.08 MJ of laser light, April 2025); eleven ignition shots to June 2026
~3,250 → 126–455
target gain a plant recirculating 10% of its output would need at NIF's efficiency, against the 5–18% efficiencies start-ups design for (The Reality Gap estimates)
55 J at 10 Hz
the largest repetition-rated diode-pumped laser run to date (Mercury); the LIFE plant design called for 2.2 MJ at 16 Hz

The governing relation

Driver efficiency and target gain are one product, and NIF's is about 80 times short of breakeven.

1101001,00010,000NIF, measured 0.7%~3,250Excimer design 5–7%~325–45510% (unit-cell target, pulser)~227Academies example 15%~150Diode-pumped design 18%~126NIF record gain 4.13target gain needed (log scale)
Target gain needed for a plant that recirculates 10% of its output (engineering gain of 10), by driver wall-plug efficiency, on a log scale; the dashed line is NIF's measured record gain of 4.13. All required gains are The Reality Gap estimates from the National Academies relation with thermal efficiency 0.4 and blanket multiplication 1.1. Not illustrative.

The National Academies define engineering gain as driver efficiency × target gain × blanket multiplication × thermal efficiency. With the Academies' 1.1 and 0.4 for the last two, the gain a target must deliver falls in direct proportion as the driver gets more efficient.

  • Wall-plug breakeven needs driver efficiency × gain of about 2.3; a plant recirculating 10% of its output needs about 23.
  • NIF's record shot scores about 0.029 (0.007 × 4.13), roughly 80 times short of breakeven and 800 short of a plant.
  • The Academies' own examples: a gain of 150 at a 15% driver and 320 at 7%.
  • Repetition rate is a separate constraint: NIF fires about once a day; plant concepts fire 1–16 times a second.

The driver register

Every efficiency above NIF's 0.7% is a design, a target or a projection.

0%5%10%15%20%NIF (measured, full scale)~0.7%Electra KrF (projected)>7%Xcimer excimer (design)5–7%Inertia unit cell (target)10%Pacific Fusion (stored-to-target)~10%LIFE / Longview (design)18%measureddesign, target or projection
Published driver efficiencies by programme, as reported in TAC-0027's register. Only NIF's is measured on a full-scale driver; Pacific Fusion's is measured against energy stored in the driver, not drawn from the wall. Hatched bars are design values, targets or projections. Not illustrative.

The register lists every inertial programme with a published driver efficiency. One row is measured at full scale. The rest describe machines that have not yet been built at plant energy or run at plant repetition rate.

  • Diode-pumped lasers (LIFE, Longview, Inertia) design for 10–18% and fast firing; every efficient joule needs pump diodes, so capital scales with average power.
  • Excimer lasers (Xcimer) accept 5–7% and plan to buy gain with very large, cheap pulses of 4–12 MJ; NRL's Electra ran 90,000 shots above 700 J and projects more than 7%.
  • Pulsed power (Pacific Fusion) delivers about 10% of stored energy to the target in its design; a one-third-scale module reached about 440 GW in June 2026.
  • Scale-up still ahead: Mercury's 55 J to LIFE's 2.2 MJ is about 40,000×; Xcimer's Phoenix (>1 kJ) to Vulcan (4–12 MJ) about 4,000–12,000×.

Three denominators

Net facility gain is a real milestone. It is not wall-plug breakeven.

Pacific Fusion measures its 2030 goal against energy stored in its capacitors. That leaves out the heat-to-electricity step, the largest single loss in any thermal plant, so a design can clear the milestone while still consuming more electricity than it returns.

  • Beryllium-liner design: simulated facility gain 1.36 → engineering gain about 0.54–0.60, a net consumer.
  • Aluminium-liner design: 4.75 → about 1.9–2.1.
  • The company's commercial target, about five times the demonstration energy: about 24 → about 9.4–10.5, near a 10%-recirculation plant.
  • Engineering breakeven needs facility gain of about 2.3–2.5; conversions assume 0.44 for thermal efficiency × blanket and 0.9–1.0 for capacitor charging (The Reality Gap estimates).
The Ecliptic, Issue No. 014, cover analysis, drawn from The Reality Gap TAC-0027 (October 14, 2026). NIF shot data, driver efficiencies, pulse energies, repetition rates and simulated facility gains are as reported in that report's cited sources; every required-gain and engineering-gain figure is a The Reality Gap estimate on the National Academies relation the report states (thermal efficiency 0.4, blanket multiplication 1.1, capacitor charging 0.9–1.0 assumed). Neither chart is illustrative.
THE ECLIPTICan Orrery publication
Content is produced by AI systems with human editorial oversight.