Industries / Frontier science / Fusion energy
Laser and inertial fusion
This is the work of crushing a tiny pellet of fuel so hard and so fast, with lasers, electrical pulses or projectiles, that it fuses before it can fly apart; nothing holds the fuel but its own inertia, which gives the method its name.
7
Milestones achieved, from 1972 to 2025.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1972
Lasers to crush fuel
Physicists at the University of California's Lawrence Livermore Laboratory published in Nature the idea that a high-energy laser could implode a pellet of heavy hydrogen to more than 10,000 times the density of the liquid, burning it efficiently enough to make fusion power plants with practical lasers feasible.
Credit
John Nuckolls, Lowell Wood, Albert Thiessen and George Zimmerman
2012
NIF misses its deadline
The National Ignition Facility (NIF) at Lawrence Livermore National Laboratory (LLNL), a laser that fires up to 192 beams at a target, ended its two-year National Ignition Campaign on 30 September 2012 without reaching ignition, the point at which the fusion energy released exceeds the energy needed to start the reaction. It also missed an intermediate step, in which particles from the fusion reactions start to add significant heat to the fuel.
LessonThe computer models used to design the shots did not match the measurements, and the US National Nuclear Security Administration pointed to that gap rather than to the laser.
Credit
Lawrence Livermore National Laboratory; reported by Physics Today
2014
Magnetized fuel on Z
At Sandia National Laboratories, the Z machine drove a current of 19 million amperes through a metal cylinder, or liner, so that the current's own magnetic field crushed it. Inside was deuterium gas threaded with a 10-tesla magnetic field and preheated by a laser; despite a slow implosion of about 70 km/s, the fuel reached about 3 keV (about 35 million degrees Celsius) and gave off up to 2 × 10^12 neutrons from fusion. This method is called magnetized liner inertial fusion (MagLIF).
Credit
M. R. Gomez, S. A. Slutz, A. B. Sefkow, D. B. Sinars and colleagues, Sandia National Laboratories
2017
Robots help build targets
The first NIF implosion targets were built by hand, and early on a rate of about one shot a week barely met demand. General Atomics, which makes NIF's targets, moved steps to a six-axis robot that applies glue, picks up a hohlraum, registers its rotation and seats it in the rest of the target, and described the cell in the peer-reviewed journal High Power Laser Science and Engineering in 2017. By its own count, automation over the decade raised its output of NIF implosion targets from one or two a week to more than one a day.
Credit
K.-J. Boehm, N. Alexander, J. Anderson, L. Carlson, M. Farrell and colleagues, General Atomics
2022
Ignition at NIF
On 5 December 2022, 2.05 MJ of laser light delivered to a target produced 3.15 MJ of fusion energy, a target gain of 1.5; the peer-reviewed paper in Physical Review Letters (February 2024) calls it the first laboratory demonstration of "scientific breakeven". By October 2026 LLNL listed 12 shots that reached ignition.
Credit
Lawrence Livermore National Laboratory and the Indirect Drive ICF Collaboration, more than 1,370 researchers from 44 institutions
2024
Direct-drive fuel gain
On the OMEGA laser at the University of Rochester, where beams strike the fuel capsule directly instead of heating a small hollow cylinder (a hohlraum) that bathes it in X-rays as at NIF, the fusion reactions gave out more energy than was held in the hot center of the fuel (a hot-spot fuel gain above 1). OMEGA did it with as little as 28 kJ of laser energy, compared with up to 1.9 MJ for the same milestone at NIF, as reported in Nature Physics.
Credit
C. A. Williams, R. Betti, V. Gopalaswamy and colleagues, Laboratory for Laser Energetics
2025
First Light changes course
First Light Fusion, a UK company founded in 2011 that pursued "projectile fusion", in which a very fast projectile strikes a target that focuses the impact onto the fuel, dropped its plans to develop a fusion power plant. It said it would instead sell its target technology, which it calls an amplifier, to other inertial fusion companies and to non-fusion users such as NASA.
LessonThe company said the change would bring revenue earlier and lower its long-term funding needs, a measure of how much money the path from experiment to power plant takes.
Credit
First Light Fusion; reported by World Nuclear News, 3 March 2025
Watch for
An 8.6-megajoule shot
LLNL reports that on 7 April 2025 NIF delivered 2.08 MJ of laser energy to a target and got back 8.6 MJ (plus or minus 0.45 MJ) of fusion energy, a target gain of 4.13, and that a shot on 20 June 2026 gave 7.9 MJ.
Achieved whenthe April 2025 shot is described in a peer-reviewed paper, as the 2022 shot was.
Claimed by
Lawrence Livermore National Laboratory, in its 2025 annual report (February 2026) and on its ignition page
A pulser built to last
LLNL and Pacific Fusion report that Sirius, a four-stage prototype pulsed-power driver, passed 3,000 shots. It is a Marx generator, a bank of capacitors charged side by side and discharged in one step, of the type Pacific Fusion plans to use to crush fuel with an electrical pulse instead of a laser; each 100-nanosecond pulse delivers 60 GW to a test load at 95% efficiency, by their numbers.
Achieved whenthe reliability data are published for outside review, and a driver of this design compresses a fusion target.
Claimed by
Lawrence Livermore National Laboratory and Pacific Fusion, reported by the American Nuclear Society, July 2026
Open
01
Shell flaws, an open benchmark
A public, labeled set of microscope images of hollow plastic or glass shells 1 to 4 mm across, the size of inertial fusion capsules, bought commercially and holding no fuel, each photographed from several sides with every surface flaw marked by two people. Add a baseline flaw detector trained on the set, and report how often it agrees with the people compared with how often they agree with each other.
The briefCredit · unclaimed02
Drivers against plant needs
A public table of every inertial fusion driver that has been built or prototyped, from flashlamp-pumped glass lasers such as NIF to diode-pumped lasers, krypton fluoride and argon fluoride gas lasers, pulsed-power machines such as Z and Sirius, and heavy-ion accelerators. For each, give the energy per shot, the shot rate, the longest continuous run and the wall-plug efficiency (the share of electricity drawn from the grid that reaches the target as driver energy), restated on one definition, beside what a power plant needs.
The briefCredit · unclaimed
Open 01 · the brief
Shell flaws, an open benchmark
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public, labeled set of microscope images of hollow plastic or glass shells 1 to 4 mm across, the size of inertial fusion capsules, bought commercially and holding no fuel, each photographed from several sides with every surface flaw marked by two people. Add a baseline flaw detector trained on the set, and report how often it agrees with the people compared with how often they agree with each other.
Why it is still open
General Atomics built a neural-network system to pick NIF capsule mandrels, the plastic shells on which capsules are built, by their surface flaws, trained on its own production images, and describes manual inspection as labor-intensive and subjective; a 2024 General Atomics poster sorts flaws on diamond capsule shells into classes such as pits and debris, again from in-house images. Inertial fusion plants would need targets in volume, because each shot destroys one. A November 2026 search found no public, labeled image set of shell flaws.
Built with what exists
A stereo microscope with a camera and a motorized rotation stage; an open labeling tool such as CVAT or Label Studio; an open vision model fine-tuned on one GPU (see the Note "One GPU to fine-tune"); Zenodo or Hugging Face to publish the set.
First step
Buy one batch of hollow polymer microspheres 1 to 2 mm across and photograph 100 of them on a rotation stage to settle the lighting and focus.
Done looks like
At least 5,000 shells imaged and labeled by two people, with flaws down to 10 micrometers marked; the set published under an open license; on held-out shells, the detector's agreement with the labels within 5 percentage points of the two people's agreement with each other.
Open 02 · the brief
Drivers against plant needs
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public table of every inertial fusion driver that has been built or prototyped, from flashlamp-pumped glass lasers such as NIF to diode-pumped lasers, krypton fluoride and argon fluoride gas lasers, pulsed-power machines such as Z and Sirius, and heavy-ion accelerators. For each, give the energy per shot, the shot rate, the longest continuous run and the wall-plug efficiency (the share of electricity drawn from the grid that reaches the target as driver energy), restated on one definition, beside what a power plant needs.
Why it is still open
The US Department of Energy's 2023 report on inertial fusion energy research needs says a plant needs about 10 shots a second and a driver whose wall-plug efficiency times the target gain is at least 10. Published figures mix definitions, such as electricity to laser light or electricity to energy on target, and Wikipedia's comparison of drivers leaves the efficiency or the shot rate blank for several of them. A November 2026 search found no maintained, sourced table on one definition.
Built with what exists
Published papers and lab reports found through the Crossref and OSTI APIs; an AI agent that pulls each figure into a row with its page citation, with a person checking every row (see the Note "The jobs agents repeat"); a spreadsheet published with dated links.
First step
Fill the rows for NIF, Sandia's Z machine and Sirius from the sources in this domain, then find the published run records of the Electra krypton fluoride laser at the US Naval Research Laboratory.
Done looks like
At least 10 drivers, each with all four figures or marked "not published", every figure linked to the page it came from; every efficiency restated on one definition with the conversion shown; each lab or company shown its row two weeks before publication, with any reply recorded.