Fusion and nuclear glossary

350 terms for confinement, fuels, gain, reactor types and licensing. Each one links to the records that use it.

Showing all 350 terms

Concepts 103

'Walk-away safe' / inherently safe: A marketing phrase, not a defined technical term, used by some reactor developers to describe designs whose passive safety features are strong enough that no operator action is needed for an extended period after certain accidents. It rests on the same underlying facts as any passive safety claim, chiefly decay heat and reactivity coefficients, and should be checked the same way: for which accidents, for how long, and against what numbers. The related engineering idea, inherent safety, is narrower: it means a particular hazard has been removed by the choice of materials or design, such as a coolant that cannot boil at the reactor's operating pressure, not that the whole plant is safe in every circumstance. The dataset records such statements as the company's own words, not as an independently established property of the design.; concept:walk-away-safe
Achieved (status): The status given to a figure that was measured on a specific run of a machine, carries a date, and has been checked by a specialist review against its published source. It is the only status eligible for the achieved leaderboards. Promoting a figure from claimed to achieved is a judgement made only during review, never at the point a figure is first extracted from a source.; concept:achieved
Actinides (plutonium and minor actinides): The heavy elements from actinium upward in the periodic table, including uranium, plutonium, neptunium, americium and curium. In reactor fuel, those heavier than uranium are made when uranium absorbs neutrons without splitting. After most fission products have decayed, actinides dominate the long-term radioactivity of spent fuel, over thousands to hundreds of thousands of years. Fast reactors can fission actinides that thermal reactors mostly cannot, which is the physical basis for claims that a reactor 'burns waste'. Such a claim is only meaningful if it says which actinides, from what fuel, at what rate, and whether the reprocessing it depends on exists. Fusion produces no actinides.; concept:actinides
Alpha heating: The heating of a fusion plasma by the helium nuclei, called alpha particles, produced by the fusion reactions themselves, as distinct from heating supplied from outside by lasers, microwaves or particle beams. A D-T reaction releases 17.6 MeV, split between a neutron carrying 14.1 MeV, which escapes the plasma and does not reheat it, and an alpha particle carrying 3.5 MeV, which is charged, stays in the plasma and deposits its energy back into the fuel. So only about a fifth of D-T fusion energy is available to keep the plasma hot. When alpha heating is the largest source of heating, the plasma is called a burning plasma; ignition is the point where alpha heating alone outweighs every way the plasma is losing energy.; concept:alpha-heating
Aspect ratio: In a magnetic confinement device, the ratio of the major radius, the distance from the machine's central axis to the centre of the plasma, to the minor radius, the radius of the plasma's own cross-section. A conventional tokamak has a relatively high aspect ratio, a wide doughnut with a thin ring of plasma: ITER's is about 3.1 (6.2 metres divided by 2.0 metres). A spherical tokamak is squeezed to an aspect ratio below about 2, closer to a cored apple, which holds plasma pressure with less magnetic field but leaves very little room at the centre for the central column and its shielding.; concept:aspect-ratio
At-the-market offering ATM offering: A way for an already-listed company to sell new shares gradually into the existing public market, at prevailing prices, rather than in one priced block like a traditional secondary offering. It lets a company raise cash incrementally without announcing a single large, dilutive round, which is why the pace and size of ATM sales can be harder for outsiders to track than a conventional funding round. No offering of this kind is currently recorded in this dataset.; concept:at-the-market-offering
Bootstrap current: A current that a tokamak plasma drives through itself, without any external current-drive system, as a side effect of the fall in pressure from its hot centre to its cooler edge. It arises from the way particles trapped on the outer, weaker-field side of the ring collide with the particles circulating freely around it. A tokamak needs a current in its plasma to confine it, and a transformer can only drive that current for a limited time, so a machine meant to run steadily must supply it some other way. The larger the share of current the plasma drives itself, the less external current drive is needed, saving both the equipment and the electricity it consumes. Advanced and spherical tokamak designs aim for a high bootstrap share.; concept:bootstrap-current
Breeder reactor: A reactor that makes more fissile fuel from fertile material, usually uranium-238 or thorium-232, than it uses up, so its stock of fissile material grows. The measure is the conversion ratio, new fissile atoms made for each fissile atom consumed: above 1 is breeding, about 1 is break-even, and below 1 is a converter, or a burner if well below. Ordinary light-water reactors make some plutonium from uranium-238 but have conversion ratios well below 1. Fast reactors are the usual route to breeding with uranium and plutonium, because at high neutron energies plutonium-239 releases more new neutrons for each neutron it absorbs, leaving more spare to convert uranium-238. Thorium is the exception that can breed with slow neutrons, narrowly: the Shippingport light-water breeder core in the United States, run from 1977 to 1982, ended with about 1.4% more fissile material than it started with. Whether a fast reactor breeds or burns is largely a design and fuel-management choice, so 'capable of breeding' does not mean a reactor is breeding, and few reactors have ever run as breeders commercially.; concept:breeder-reactor
Bremsstrahlung (radiation loss): X-rays given off when electrons in a plasma are deflected by ions. Every hot plasma loses energy this way, and the loss grows with the electric charge of the ions present, so impurities knocked off the wall, and fuels containing heavier nuclei such as boron, radiate far more than pure hydrogen fuel. It is a main reason proton-boron-11 fusion is so much harder than D-T: at the temperatures p-B11 needs, the energy radiated comes close to the fusion power the plasma produces, leaving very little margin.; concept:bremsstrahlung
Burning plasma: A plasma in which heating by the fusion reactions' own alpha particles is the largest source of heating, outweighing the heating supplied from outside. ITER describes it as a plasma in which self-heating dominates any other form of heating. The National Ignition Facility reported reaching it in a Nature paper published in January 2022, from experiments that came before its first target gain above 1. In a magnetic device, alpha heating overtakes external heating once plasma gain passes 5, since the alpha particles carry a fifth of D-T fusion energy; no magnetic device has reached that, and ITER's design gain of 10 would put it well inside the regime. A burning plasma is a step towards ignition, not the same thing.; concept:burning-plasma
Burnup: A measure of how much energy has been extracted from nuclear fuel before it is removed from the reactor. It is usually stated in gigawatt-days per tonne of heavy metal (GWd/t), where heavy metal means the uranium or other actinides in the fuel, or as the percentage of heavy metal atoms that have fissioned. Megawatt-days per tonne (MWd/t) is a unit a thousand times smaller, and mixing the two up is a common error. Light-water reactor fuel is typically removed at roughly 40 to 60 GWd/t, about 5% of its heavy metal atoms fissioned. Higher burnup means fewer refuellings and less spent fuel by mass for the same energy, although each tonne of it is more radioactive. Every fuel form is qualified only up to the burnup it has been tested to: the US Department of Energy reports that TRISO particles reached a peak of 19% of heavy metal atoms fissioned in a 2009 test, about three times what current light-water fuel achieves.; concept:burnup
Campaign: A planned run of experiments on a device, usually built around a specific fuel, configuration or goal, made up of many individual shots or pulses. JET's third and final deuterium-tritium campaign, in 2023, included the record 69.26 megajoule pulse 104522. A device's best result for a given metric may come from any shot or pulse within a campaign, and a result should be dated to that shot, not to the end of the campaign or the announcement, wherever the source allows.; concept:campaign
Capsule (fuel capsule): The small shell of fuel, a few millimetres across, that is compressed in inertial confinement fusion until it fuses. The fuel is a thin layer of frozen deuterium-tritium on the inside of the shell. In laser-driven fusion the capsule is imploded either by laser light striking it directly or, at the National Ignition Facility, by X-rays generated inside a surrounding hohlraum. The target is the whole assembly, hohlraum and capsule together. How much energy the capsule actually absorbs, as opposed to the laser energy delivered to the whole target, is the difference between capsule gain and target gain.; concept:capsule
Capture: The first step in this site's data pipeline: fetching a source page and storing an exact copy of what came back, along with a SHA-256 fingerprint of the stored bytes. Every later check is made against this stored copy, not the live page, so if the page is later edited or removed, what this site relied on cannot silently change. A page that loads successfully but sends little readable content, because scripts were meant to fill it in afterward, is marked thin and re-fetched with a full browser rather than trusted as captured.; concept:capture
Central solenoid: The large coil through the middle of a tokamak that works as a transformer, with the plasma ring acting as its secondary winding: changing the coil's current induces the current in the plasma and also heats it. A transformer can only keep changing its current for so long, so a tokamak run this way is pulsed. Running without it, called non-inductive current drive, relies on bootstrap current and external current-drive systems, and is what a 'fully non-inductive' result means. Spherical tokamaks have very little room at the centre for a solenoid, which makes other ways of driving current more pressing for them.; concept:central-solenoid
Chain reaction: The process by which neutrons released from one fission cause further fissions, which release more neutrons in turn. Reactor physicists track its size, generation to generation, with the effective multiplication factor, k: below 1 the reaction dies away (subcritical), at 1 it holds steady (critical), above 1 it grows (supercritical). Every operating power reactor runs critical essentially all the time; 'critical' is the normal operating state, not an emergency, despite how news coverage sometimes uses 'went critical'.; concept:chain-reaction
Circular sourcing: The situation where a claim appears to be confirmed by multiple sources, but every one of them traces back to the same original announcement, or even back to this site's own material being copied by an aggregator and then found again in a later search. Repetition is not corroboration: a company press release repeated by a dozen outlets in a day is still one source, the company. Checking for circular sourcing means finding the earliest version of a claim and asking who actually generated it, rather than counting how many places report it.; concept:circular-sourcing
Claimed (status): The status given to a figure that is reported by a source but has not been independently confirmed. It is the status of most figures in this dataset. Claimed does not mean doubted: it means the source says so and the figure has not yet been checked against a second, independent source or promoted by a specialist review. Nothing is ever entered directly as achieved; every new figure starts as claimed.; concept:claimed
Combined licence (COL) COL: A single US Nuclear Regulatory Commission application, under the newer Part 52 framework, that covers both a construction permit and an operating licence in one proceeding, rather than the two separate applications the older Part 50 route requires. Oklo's Aurora reactor was pursued through a combined licence application, which the NRC denied without prejudice in January 2022 for missing information. Holding a combined licence application does not mean either permission has been granted; it means both are being sought together.; concept:combined-licence
Confidence: A field recording this site's judgement of how well the evidence supports a given record, at one of three levels: high, medium or low. Anything resting on a fact added from outside a captured source, rather than taken directly from one, enters at low confidence. Confidence is a judgement about the evidence, not a claim about whether the underlying fact is true.; concept:confidence
Control rods: Rods containing strong neutron absorbers, such as boron or hafnium, that operators insert or withdraw to change a reactor's reactivity: inserting them lowers the chain reaction's growth rate, withdrawing them raises it. A full, fast insertion to shut the reactor down is called a scram or trip. Several small reactor designs use rotating control drums with an absorbing face instead of rods, and pressurised water reactors also adjust boron dissolved directly in the coolant for slower, longer-term control.; concept:control-rods
Coolant: The substance that carries heat out of a reactor core to where it can raise steam or otherwise generate power. In a pressurised water reactor, ordinary water is both the coolant and the moderator; other designs separate the two jobs, cooling with helium, carbon dioxide, liquid sodium, liquid lead or molten salt while moderating with graphite or nothing at all. A reactor's coolant is not the same thing as its fuel form: a salt-cooled reactor with solid fuel and a molten-salt reactor that dissolves its fuel in the salt are different designs that happen to share a coolant.; concept:coolant
Criticality: The state of a nuclear chain reaction in which each generation of fissions is exactly as large as the one before it, so the reaction is self-sustaining at a steady rate rather than dying away or growing. A reactor operating at steady power is critical continuously; this is its normal state. To raise power, operators make it slightly supercritical for a while, and to lower power, slightly subcritical. First criticality, the first time a specific reactor achieves this, is a distinct commissioning milestone, tracked in this dataset as a regulatory stage.; concept:criticality
Decay heat: The heat produced by the radioactive decay of fission fragments inside reactor fuel, which continues long after the chain reaction itself has been shut down. Immediately after shutdown it is about 6.5% of the power the reactor was running at, falling to roughly 1 to 1.5% after an hour, about 0.4% after a day and 0.2% after a week; the exact figures depend on how long the reactor ran and on its fuel. For a large reactor producing 3,000 MW of heat, that is about 195 MW at shutdown and still about 12 MW a day later. Losing the ability to remove decay heat, not a runaway chain reaction, caused the fuel damage at Three Mile Island and Fukushima Daiichi. Decay heat scales with reactor power, while the surface through which heat can leak away shrinks more slowly than the volume as a reactor gets smaller, so a very small reactor can plausibly shed its decay heat to its surroundings by conduction and natural airflow. Large reactors can also remove it passively for a stated period, but only through dedicated systems such as raised water tanks and natural circulation loops.; concept:decay-heat
Delayed neutrons: A small fraction of the neutrons from fission that are not released at the instant of fission but a fraction of a second to about a minute later, as certain fission fragments decay radioactively. For uranium-235 this fraction, written β (beta), is about 0.64% of all fission neutrons. It is markedly smaller for plutonium-239, about 0.2%, and for uranium-233, about 0.26%, so a core fuelled mainly with either has less margin to work with. Without delayed neutrons, the neutron population in a thermal reactor would change on a timescale of about a ten-thousandth of a second, and far faster in a fast reactor, too fast for control rods or operators to follow. Because reactors are run so that they need the delayed fraction to stay critical, the effective time between neutron generations stretches to roughly a tenth of a second, which is what makes a reactor controllable at all. A reactor that could sustain its chain reaction on prompt neutrons alone is called prompt critical, a state normal operation avoids by a wide margin.; concept:delayed-neutrons
Denied without prejudice: A regulator's denial that leaves the applicant free to reapply once the missing information or deficiency is fixed, as opposed to a denial that closes the matter permanently. The NRC's January 2022 denial of Oklo's Aurora combined licence application was without prejudice, because the agency said Oklo had not provided information it had requested. A denial without prejudice is still a real denial and a first-class milestone in this dataset; it is not the same thing as an application merely being paused.; concept:without-prejudice
Denominator: What a ratio, most often a gain figure, is divided by. Fusion gain is meaningless without stating it: the same NIF shot in December 2022 was a target gain of 1.5 when its fusion energy is divided by the laser energy delivered to the target, and a wall-plug gain below 0.01 when it is divided by the electricity the facility stored to fire the shot. This site records a denominator for every ratio it publishes, which is why it has no field that can hold a bare gain figure at all.; concept:denominator
Depleted uranium (tails): The uranium left over from enrichment, holding less uranium-235 than natural uranium, typically 0.2 to 0.3%. It builds up in large quantities: making one kilogram of ordinary 4.5% reactor fuel leaves roughly eight kilograms of it. It is less radioactive than natural uranium and is mostly stored. It is the uranium usually blended with plutonium in MOX fuel, and the fertile material a fast reactor could convert into plutonium.; concept:depleted-uranium
Design value: A figure describing what a machine was built to do, such as its major radius, planned magnetic field or intended plasma current, rather than a measured result of running it. Design values are shown separately from performance results, most prominently on the devices pages, and are never mixed into the achieved or claimed performance tables: a design target and a measured shot answer different questions, even when they use the same units.; concept:design-value
Direct energy conversion: Turning the energy of charged fusion products, or of an expanding magnetised plasma, straight into electricity, without first making heat and running a steam or gas turbine. It works only for energy carried by charged particles, so it suits fuels such as deuterium-helium-3 and proton-boron-11, whose main reactions release few neutrons. D-T puts about 80 per cent of its energy into neutrons, which can only be collected as heat. Helion Energy says it will recover electricity from the plasma's expansion in each pulse. The dataset holds no measurement of electricity produced this way, or any other way, by a fusion device.; concept:direct-energy-conversion
Disputed (status): The status given to a figure for which a credible challenge exists, such as a second source giving a materially different account of the same result. A disputed figure remains in the record with the dispute noted, rather than being resolved by picking one side, because the disagreement itself is part of the evidence.; concept:status-disputed
Disruption: A sudden loss of control of a tokamak plasma. Its stored heat is dumped onto the wall within milliseconds, then the plasma current collapses, inducing large forces in the vessel and magnets and sometimes producing beams of very fast 'runaway' electrons that can damage the wall. Disruptions are one of the main open engineering problems for tokamaks: a reactor-scale machine must predict and avoid them, or survive them, because neither the forces nor the heat loads can be allowed on a wall built to last for years. Stellarators do not need a current in the plasma to confine it, so they are largely free of this failure mode.; concept:disruption
Docket: The formal case file a regulator opens once it accepts an application for review, identified by a docket number that can be used to look up the regulator's own record of every filing and decision on that application. In United States nuclear licensing, an application being 'accepted for docketing' means the paperwork is complete enough to examine, not that it has been approved; the docket number is what this site's licensing page records as the anchor for a milestone, so a reader can verify a claim directly against the regulator's own record.; concept:docket
Engineering breakeven: The point where a fusion plant generates as much electricity as it consumes, counting everything the plant uses to run itself: magnets, cooling, controls and heating systems included. This is an engineering gain of 1; a plant that sells electricity needs more than 1. No fusion device has produced electricity at all, so the engineering gain of every device built so far is zero, including those that have passed scientific breakeven.; concept:engineering-breakeven
Enrichment: The industrial process of raising the share of the fissile isotope uranium-235 in a batch of uranium above its natural abundance of about 0.7%, usually by spinning uranium hexafluoride gas in centrifuges. The enrichment level decides which category the fuel falls into. Anything below 20% uranium-235 is low-enriched uranium; within that, this site separates conventional fuel up to 5%, LEU+ at 5 to 10% and HALEU at 5% to just under 20%. Uranium at 20% or more is highly enriched. The effort needed is measured in separative work units and rises steeply with the target level. Enrichment is the chokepoint of the fuel cycle: it needs specialised, tightly controlled facilities, and Western capacity for the HALEU band is currently thin relative to the demand advanced reactor designs are creating.; concept:enrichment
External heating: The ways energy is put into a plasma from outside: driving a current through it, which heats it the way current heats a kettle element (ohmic heating); firing beams of fast neutral atoms into it (neutral-beam injection); or beaming in radio waves or microwaves tuned to the ions or the electrons. In laser fusion the laser plays this role. The heating power that reaches the plasma is the denominator of plasma gain. The electricity these systems draw is larger, because none of them is perfectly efficient, and only engineering and wall-plug gain count it. Sources also differ on whether they count power injected or power absorbed, which is one reason one JET shot from 1997 is quoted as both 0.63 and 0.67.; concept:plasma-heating
First plasma: The first time a newly built fusion device successfully forms a plasma at all. It demonstrates that the machine works as an integrated system, magnets, vacuum, heating and control together, but says nothing about performance: temperature, density and confinement time are usually far below what the same machine will eventually reach. First plasma is the earliest rung on the fusion milestone ladder, well before scientific breakeven or ignition.; concept:first-plasma
First wall: The innermost surface of a fusion device: the first solid material the plasma's heat and particles reach. It must survive constant particle bombardment, intense heat and, in a D-T machine, a steady flux of 14.1 MeV neutrons, all at once and for years at a time. No first-wall material has yet been tested at the neutron dose a power plant would impose, because no neutron source of the right energy and intensity to do so yet exists.; concept:first-wall
Fissile versus fertile: A fissile nucleus, such as uranium-235 or plutonium-239, can be split by a neutron of any energy, including a slow one, and so can sustain a chain reaction. A fertile nucleus cannot sustain a chain reaction itself, but after absorbing a neutron it turns, through two radioactive decays, into a fissile one. Uranium-238 and thorium-232 are the two fertile materials that matter in practice, becoming plutonium-239 and uranium-233 respectively. Uranium-238 can be split by fast neutrons, and supplies a small share of the power in an ordinary reactor this way, but not often enough to keep a chain reaction going on its own. That is why a thorium reactor still needs a starting charge of fissile fuel.; concept:fissile-vs-fertile
Fission products: The lighter nuclei left when a heavy nucleus splits, usually two per fission, spread across many elements such as caesium, strontium, iodine and xenon. Most are radioactive, and their decay is what produces decay heat after a reactor shuts down. They supply most of the radioactivity of spent fuel for its first few hundred years; caesium-137 and strontium-90, each with a half-life of about 30 years, dominate that period. Some, such as xenon-135, absorb neutrons strongly and affect how quickly a reactor can change power. Keeping fission products contained, inside the fuel, its cladding and the containment building, is most of what reactor safety is about.; concept:fission-products
Fuel qualification: The body of irradiation tests, examination and analysis showing that a fuel design behaves as predicted over a stated range of temperature, burnup and power, which a regulator accepts as a basis for licensing. It usually takes many years, because the fuel has to be irradiated in test reactors for long periods and then examined. A fuel is qualified for a particular envelope, not in general: a claim that it performs beyond the conditions it was tested under is not supported by that qualification, and a new fuel form, enrichment or manufacturer usually needs its own testing.; concept:fuel-qualification
Fusion milestone ladder: The sequence of steps commonly used to describe fusion progress, in order: first plasma, running the real fuel (deuterium-tritium) at scale, scientific breakeven, ignition, engineering breakeven, net electricity delivered to the grid, and commercial operation. The ladder is a simplification. In laser fusion, a burning plasma, ignition and scientific breakeven came within about two years of each other, and ignition and breakeven fell on the same shot; a magnetic-confinement power plant need not reach full ignition at all, only a high enough plasma gain. Different steps have been reached by different devices using different fuels, so no single device has climbed the whole ladder, and a company's roadmap should be read against which specific rung it is claiming.; concept:fusion-milestone-ladder
Generic Design Assessment (GDA) GDA: A voluntary UK process, run jointly by the Office for Nuclear Regulation and the Environment Agency (and Natural Resources Wales in Wales), that assesses a reactor design on its own merits, separately from any specific site. It proceeds in steps that get more detailed, normally three, and a full GDA ends with a Design Acceptance Confirmation and a Statement of Design Acceptability. Even a completed GDA does not authorise building anything: the operator still needs a nuclear site licence, environmental permits and planning permission for a specific site.; concept:generic-design-assessment
Gross versus net output: A power plant's gross output is the electricity its generator produces; its net output is what is left for the grid after the plant's own use, such as pumps, fans, cooling and control systems. The gap is usually a few percent for a large plant and can be wider for a small one, so a rating that does not say which it means cannot be compared exactly with one that does. The IAEA's 2024 catalogue, for example, lists Holtec's SMR-300 at 366 MWe gross and 320 MWe net. This site records a rating as the developer states it and does not convert between the two.; concept:gross-vs-net-output
HALEU supply constraint: The shortage of Western capacity to produce high-assay low-enriched uranium, the 5% to just under 20% uranium-235 fuel many advanced reactor designs are built around. The World Nuclear Association names Russia's Tenex, part of Rosatom, as the only commercial supplier, and the war in Ukraine, sanctions and the US ban on Russian uranium imports in May 2024 turned that from a cost issue into a strategic supply risk for Western developers. Several governments and companies have since funded new enrichment capacity, but building and licensing a new enrichment line, and qualifying the fuel made from it, takes years, so the constraint affects how plausible a near-term timeline is for any design that assumes HALEU fuel. It is a shortage of plants, not of uranium.; concept:haleu-supply-constraint
Half-life: The time it takes for half of a given amount of a radioactive isotope to decay. It is not a measure of danger on its own: a short half-life means the atoms decay quickly, so a given amount is intensely radioactive for a short time, while a long half-life means weaker radioactivity that lasts longer. Tritium's half-life of 12.32 years means a stock loses about 5.5% a year to decay, which, together with a small world supply, is why a deuterium-tritium fusion plant must breed its own. The six standard groups of delayed neutron emitters in fission have half-lives from about 0.23 to 55.7 seconds, which is what slows a fission reactor's response enough for control rods and operators to keep up.; concept:half-life
High-confinement mode (H-mode) H-mode: An operating regime of magnetic confinement devices, mostly tokamaks, entered when the heating power passes a threshold. A thin layer at the plasma edge suddenly suppresses turbulence, building a steep step in pressure called the pedestal, and the plasma holds its heat roughly twice as well as in L-mode on the same machine. It was first reported on the ASDEX tokamak in Germany in 1982 (Wagner and colleagues, Physical Review Letters). H-mode is the baseline operating mode for ITER and most large tokamak designs. It has a cost: the steep edge collapses repeatedly in bursts called edge-localised modes (ELMs), which throw heat and particles onto the wall and must be controlled in a reactor. A plasma described as a long 'high-confinement' plasma is being described by its operating mode and its duration; its energy confinement time is still typically a second or less.; concept:h-mode
Highly enriched uranium (HEU) HEU: Uranium enriched to 20% uranium-235 or more. The 20% line is the international boundary between low-enriched and highly enriched uranium, and material above it is subject to much stricter security and safeguards controls. That is why many advanced reactor designs specify 19.75%: as much uranium-235 as possible while staying below the line. Highly enriched uranium is used in naval propulsion and some research reactors, and international programmes have spent decades converting research reactors to low-enriched fuel.; concept:highly-enriched-uranium
Hohlraum: A small metal can, about a centimetre long and usually made of gold or depleted uranium, used in indirect-drive laser fusion. Lasers heat the inside of the hohlraum until it glows with X-rays, and those X-rays, rather than the laser light itself, implode the fuel capsule held at its centre. The National Ignition Facility uses this method. Converting laser light to X-rays and then into the capsule costs energy: roughly a tenth of the laser energy delivered to a hohlraum ends up absorbed by the capsule inside it. That is why capsule gain is so much larger than target gain. Direct drive avoids the conversion but must light the capsule far more evenly.; concept:hohlraum
How fusion is regulated: Fusion devices are generally not licensed under the same rules as fission reactors, because they hold no self-sustaining chain reaction and carry a different, much smaller radiological hazard. In the United States, the Nuclear Regulatory Commission decided in April 2023 to regulate fusion facilities under its rules for radioactive materials rather than its reactor licensing framework, with definitions written into law by the ADVANCE Act in 2024. In the United Kingdom, the Energy Act 2023 states that a fusion energy facility does not need a nuclear site licence, except where it shares a site with a fission reactor. Because of this split, this site's regulatory stage vocabulary is applied only to fission facilities.; concept:fusion-regulation
Ignition: The point where a fusion plasma's own reactions supply enough heat to keep it hot, with no more heating from outside needed. For laser-driven fusion, Lawrence Livermore National Laboratory defines it physically, as the point where heating by the fusion reactions' own alpha particles overcomes the fuel's cooling losses, and says the National Ignition Facility reached it on 5 December 2022. That shot was also the first to release more fusion energy than the laser energy delivered to the target, the simpler test the US Department of Energy used. NIF's peer-reviewed papers report that the physical threshold had already been crossed on a shot in August 2021 that released less energy than its laser delivered, so the two definitions give different dates. In magnetic confinement, ignition would mean a plasma needing no external heating at all, which no device has reached, and a power plant would not need to reach it: a high enough plasma gain is sufficient. The word is also used loosely in the press for any energy-positive result, so check which definition a claim uses.; concept:ignition
Implosion (inertial fusion): The inward collapse of an inertial fusion capsule. The laser light or X-rays blast material off the capsule's outer surface, and, like a rocket exhaust, this drives the rest of the shell inward at hundreds of kilometres per second. The fuel is compressed to far higher density than any solid, and a small region at the centre, the hot spot, is heated enough to start fusing. The implosion must be almost perfectly even: one that goes lopsided loses much of its compression, and with it most of its yield.; concept:implosion
Isotope: One of two or more forms of the same chemical element whose atoms have the same number of protons but a different number of neutrons, giving them the same chemistry but different mass and, often, different nuclear behaviour. Deuterium and tritium are isotopes of hydrogen with one and two neutrons, where ordinary hydrogen has none; uranium-235 and uranium-238 are isotopes of uranium that behave very differently in a reactor, the first readily fissile, the second not. Enrichment is the process of raising the share of one particular isotope, uranium-235, in a batch of uranium.; concept:isotope
Kiloelectronvolt (keV) keV: A unit of energy used to state plasma temperature, equal to one thousand electronvolts. Plasma physicists prefer it to kelvin or Celsius because it describes the energy of the particles directly: one keV is about 11.6 million kelvin, so a headline '100 million degree' plasma is about 8.6 keV. Energies released by single fusion reactions are given in the larger unit MeV, a million electronvolts: one D-T reaction releases 17.6 MeV. Ion temperatures in this dataset are given in keV, with a kelvin conversion shown alongside on the leaderboards.; concept:kev
Laser drive: Using laser light as the energy source that compresses an inertial-confinement fusion target, either striking the fuel capsule directly (direct drive) or heating a surrounding hohlraum that then compresses the capsule with X-rays (indirect drive). The laser's own efficiency, how much electricity it takes to put a given amount of light on target, multiplies directly into wall-plug gain. NIF, built for research rather than efficiency, stores about 400 MJ of electricity to deliver about 2 MJ of light, roughly half a per cent. A power plant needs a far more efficient laser and a far higher target gain together: the laser efficiency multiplied by the target gain and by the efficiency of turning fusion heat into electricity must comfortably exceed 1 before the plant returns any electricity.; concept:laser-drive
Lawson criterion: A calculation made in 1955 by the British physicist John Lawson of the plasma conditions a fusion reactor must meet to produce more energy than it loses. It was classified at first and published in 1957. In its original form it is a threshold on density multiplied by energy confinement time, which depends on temperature; folding temperature in gives the triple product used throughout this benchmark. The version most quoted today is the condition for ignition, where the fusion reactions' own heating balances every loss: for deuterium-tritium fuel, a triple product of about 3 × 10²¹ keV·s·m⁻³ at an ion temperature of about 14 keV. A plasma short of that threshold can still give useful gain.; concept:lawson-criterion
Leaderboard eligibility: The rules that decide whether a figure appears on a ranked leaderboard rather than only on the benchmark tables. Planned, superseded, disputed and retracted figures are excluded, as are design parameters, which describe what a machine was built to do rather than what it did, and figures whose unit cannot be converted to the board's common unit. A board is shown only once at least three devices have a qualifying figure, and each device appears once, at its best value.; concept:leaderboard-eligibility
Ledger: An append-only log of every change made to the dataset, in which each entry carries a fingerprint of the entry before it, so that altering or deleting a past entry would break the chain and be detectable by a routine check. Records are never deleted to resolve a problem; instead their status is lowered and a note explains why, and the ledger preserves that history. The site's public changelog is built from this ledger.; concept:ledger
Light-water reactor LWR: The family of reactors that use ordinary water as both coolant and moderator: pressurised water reactors and boiling water reactors. Together they make up about 85% of the world's operable power reactors, 376 of 441 units in the World Nuclear Association's count of 5 August 2026. Because ordinary water absorbs some neutrons, they need uranium enriched to a few percent uranium-235. 'Light' distinguishes ordinary water from heavy water, whose hydrogen is deuterium.; concept:light-water-reactor
Low-confinement mode (L-mode) L-mode: The ordinary operating regime of a magnetically confined plasma, without the edge pedestal and roughly doubled confinement of H-mode. Every tokamak runs in L-mode at low heating power and enters H-mode only above a heating threshold. Not every plasma that is not in H-mode is in L-mode: other improved regimes exist, such as the 'supershots' used for many of TFTR's best results and plasmas with a barrier to heat loss deep inside them (an internal transport barrier). The regime of a figure should therefore come from its source, not be assumed.; concept:l-mode
Magnetic mirror losses: The escape of plasma particles out of the ends of a magnetic mirror device. A mirror holds plasma in a straight tube whose magnetic field is stronger at each end, and a particle heading towards an end is reflected back only if enough of its motion is spent circling the field line rather than travelling along it. Particles moving too directly along the axis fall inside what is called the loss cone and pass straight out through the strong-field end. Collisions keep knocking new particles into the loss cone, so the leak never stops. Reducing it, with stronger end fields, extra plugging cells at the ends or other tricks, has been the central problem of mirror fusion since the 1950s.; concept:magnetic-mirror-losses
Moderator: A material used in many reactors to slow down, or moderate, the fast neutrons released by fission, because uranium-235 is far more likely to fission when struck by a slow neutron. Ordinary water slows neutrons in the fewest collisions but absorbs some of them; heavy water absorbs far fewer, which is why a heavy-water reactor can run on unenriched uranium; graphite absorbs few neutrons but needs more collisions to slow them, so graphite cores are large. A reactor with no moderator at all runs on fast neutrons instead, and is called a fast reactor.; concept:moderator
Modular construction: Building a reactor largely from factory-made modules sized to travel by road, rail or sea, then assembling them on site, instead of building most of it on site as a one-off. In 'small modular reactor' the word also often means a plant made of several identical reactor units that can be added over time. The economic case for SMRs rests on factory production and repeated builds bringing costs down. No SMR design has yet been built as a series of factory-made units, so that case is a projection, not a demonstrated result.; concept:modular-construction
MWe versus MWt: The distinction between a reactor's electrical output, MWe (megawatts electric), and its thermal output, MWt or MW(th) (megawatts of heat produced by the core). For a conventional steam-cycle plant, electrical output is typically about a third of thermal output, because most heat engines cannot convert all their heat into electricity. Quoting a design's thermal rating where a reader expects an electrical one inflates the apparent size of a reactor without stating anything false, so it is worth checking which figure a headline is using, especially for microreactors, several of which developers describe in MWt rather than MWe.; concept:mwe-vs-mwt
Natural circulation: Coolant flow driven by gravity and temperature difference alone: coolant heated in the core becomes less dense and rises, cooler coolant sinks to take its place, and the loop keeps moving with no pump. Many passive cooling systems rely on it, and some reactor designs, such as NuScale's and the BWRX-300, use it to cool the core in normal operation too. How much heat it can carry depends on the height between where heat is made and where it is removed, and on the flow paths staying clear, so a design that relies on it should state how much heat it can remove, not only that it uses it.; concept:natural-circulation
Net energy: A phrase used loosely in fusion coverage to mean any result where more energy came out than went in, without saying which energy was counted in. Most often it means scientific breakeven, energy out exceeding the energy delivered to the fuel or plasma. It can also mean engineering breakeven, a whole plant's electricity output exceeding its electricity consumption, and the two differ enormously: NIF's December 2022 shot passed the first with a target gain of 1.5, while its wall-plug gain was below 0.01 and it produced no electricity at all. A claim of 'net energy' should always be checked against which of the six gain figures it refers to.; concept:net-energy
Neutron: An electrically neutral particle found in every atomic nucleus except ordinary hydrogen, alongside protons. Fission releases neutrons that can go on to split further nuclei, sustaining a chain reaction; fusion of deuterium and tritium releases a neutron carrying 14.1 of the reaction's 17.6 million electronvolts, which escapes the plasma and must be captured to make heat or to breed tritium. Because neutrons carry no charge, they pass easily through matter that would stop a charged particle, which is what makes them useful for breeding fuel and damaging to materials at the same time.; concept:neutron
Neutron activation: The process by which a material struck by neutrons becomes radioactive itself, because some of its atoms absorb a neutron and turn into an unstable isotope. Fusion neutrons and fission neutrons both activate the structures around them; the resulting waste is a different problem from spent fuel, since it is the reactor's own materials rather than the fuel that becomes radioactive. Choosing structural materials that activate less, or decay faster, is a major line of nuclear materials research for both fission and fusion.; concept:neutron-activation
Neutron spectrum (thermal and fast): The range of speeds, and so energies, of the neutrons doing the fissioning in a reactor. A thermal reactor uses a moderator to slow its neutrons down to roughly the thermal energy of the surrounding material before they cause fission; a fast reactor uses no moderator and fissions with neutrons close to the energy they were born at. The spectrum decides a great deal downstream: it changes how enriched the fuel must be, which elements the reactor can fission or breed, and how much damage the neutrons do to structural materials, so a claim made about a thermal design cannot be assumed to hold for a fast one, or the reverse.; concept:neutron-spectrum
Nuclear fuel cycle: The chain of steps that supplies a fission reactor with fuel and deals with what comes out. The front end is mining uranium, converting it to uranium hexafluoride gas, enriching it and fabricating it into fuel. The back end is storing spent fuel, then either reprocessing it or disposing of it. A once-through cycle, used by most countries, stores spent fuel for eventual disposal; a closed cycle reprocesses it to recover uranium and plutonium for new fuel. Each step needs its own specialised, licensed facility, so a new reactor design depends on every step being able to supply its particular fuel, not only on the reactor itself.; concept:nuclear-fuel-cycle
Nuclear site licence (UK): The licence the Office for Nuclear Regulation grants to the operator of a specific site in Great Britain before a nuclear installation can be built or run there, with conditions that apply for the life of the site. A completed Generic Design Assessment does not provide one: the design is assessed without a site, and a site licence, environmental permits and planning permission all have to be obtained separately. Under the Energy Act 2023, a fusion energy facility does not need a nuclear site licence unless it shares a site with a fission reactor.; concept:nuclear-site-licence
Part 53 (US advanced reactor licensing) Part 53: A new, optional licensing framework published by the US Nuclear Regulatory Commission in March 2026 (10 CFR Part 53), in force from 29 April 2026, that sets safety standards applicable to any reactor technology rather than being written specifically around water-cooled reactors. It adds a third licensing route alongside the older Part 50 (construction permit, then separate operating licence) and Part 52 (combined licence) frameworks; it does not remove the separate design, construction and operating steps those routes require.; concept:part-53
Passive safety: A safety feature that does its job using physics alone (gravity, natural circulation of hot and cool coolant, conduction of heat, or reactivity feedback) with no outside power, pumps or operator action. Some systems described as passive still need one active step to start them, such as a battery-powered valve opening, and that difference is worth checking. A claim of passive safety is always a claim about a specific set of accidents and usually a specific length of time, such as three days with no action needed; it is not a claim of safety against every possible event indefinitely. It depends on numbers that a reader is entitled to ask for: chiefly the reactor's decay heat, the path that heat takes out of the core, and its reactivity coefficients. A design being passively safe does not remove the need for regulation, maintenance or security around the radioactive material it still holds.; concept:passive-safety
Peak and sustained: The qualifier this site attaches to temperature and duration figures. A peak value is the highest reached, possibly for only milliseconds or microseconds; a sustained value was held for a stated length of time; an average is taken over a stated period. A pinch that reaches a high temperature for a few microseconds and a tokamak that holds a lower temperature for 48 seconds have done different things, and neither simply beats the other. 'Peak' is also used in a second sense, for the value at the hottest point in space, usually the centre of the plasma, rather than an average over its volume, so a source's own wording should be read closely.; concept:peak-vs-sustained
Pinch: A family of approaches that confine a plasma using the magnetic field generated by a current flowing through the plasma itself, which squeezes, or pinches, the plasma column inward. A Z-pinch drives the current along the length of a straight column and needs no external confining coils. A reversed-field pinch is a ring-shaped variant with a weaker external field whose direction reverses near the plasma edge. A pinched column tends to kink sideways or neck into sausage-like segments, and each variant suppresses these instabilities differently: Zap Energy, for example, runs a flow of plasma along the column at different speeds at different radii. Pinch plasmas typically last microseconds.; concept:pinch
PIPE (private investment in public equity) PIPE: A financing deal in which investors buy shares of a company directly, at a negotiated price, after it is already publicly listed, rather than through a public share sale on the open market. PIPE deals are often used alongside a SPAC merger to provide additional capital at the time a company goes public that way. No deal of this kind is currently recorded in this dataset; if one is captured, it should be added to the funding-instrument vocabulary rather than tracked only here.; concept:pipe
Planned (status): The status given to a target, design goal or roadmap date rather than a measured result. A planned figure is shown on this site, but it is permanently excluded from every achievement leaderboard, because a target is not a result, however precisely stated. ITER's projected plasma gain of 10 and Commonwealth Fusion Systems' projected gain of about 11 for SPARC are both held as planned.; concept:planned-figure
Plasma: A gas heated so hot that electrons separate from their nuclei, leaving a mix of charged particles that behaves differently from an ordinary gas: it conducts electricity and responds strongly to magnetic and electric fields. Every fusion fuel becomes a plasma before it can fuse. The Sun, lightning and the inside of a fluorescent tube are all plasmas, at very different temperatures and densities.; concept:plasma
Plasma beta β: The ratio of a plasma's pressure to the pressure of the magnetic field that confines it (the field's 'magnetic pressure', which rises with the square of the field strength). A higher beta means the magnetic field is being used more efficiently to hold a given plasma pressure, and because fusion power rises steeply with plasma pressure, it generally means a smaller or cheaper magnet for the same fusion power. Configurations differ sharply: conventional tokamaks run at a few per cent, spherical tokamaks have reached tens of per cent, and field-reversed configurations run close to 100 per cent. Beta is defined in more than one way, using different parts of the field, so compare two figures only if they use the same definition.; concept:beta
Plasma instability: Any process by which a small disturbance in a plasma grows instead of dying away. The large-scale ones, often called MHD (magnetohydrodynamic) instabilities, bend or break the whole plasma: a pinched column can kink sideways or neck into sausage-like segments, and a tokamak plasma pushed too far in current, density or pressure can disrupt. Smaller-scale turbulence does not destroy the plasma but carries heat out of it faster than simple theory predicts, and largely sets the energy confinement time. Much of fusion research since the 1950s has been about understanding and suppressing both.; concept:plasma-instability
Provenance: The record, attached to every claim on this site, of exactly which source or sources it came from. A claim with no provenance is treated as unsupported and fails the site's validation checks; provenance is what lets a reader trace any figure back to the document it was taken from, rather than trusting the site's summary on faith.; concept:provenance
Pulse: One period of plasma operation on a magnetic confinement device, from the plasma forming to it ending, identified by a pulse number where the source gives one. JET's record pulse, number 104522 on 3 October 2023, released 69.26 megajoules of fusion energy over five to six seconds (UKAEA says five, EUROfusion six). Pulse and shot mean the same thing; which word a laboratory uses is a matter of habit. The length of a pulse is its plasma duration, which is not the same as its energy confinement time.; concept:pulse
Q (fusion gain, use with caution) Q: The letter commonly used in coverage and by companies for fusion gain, energy out divided by energy in. It is ambiguous on its own, because a fusion experiment spends energy at several different stages and each stage gives a different answer: the same NIF shot in December 2022 was a target gain (Q_target) of 1.5 and a wall-plug gain below 0.01. By convention Q usually means plasma gain in magnetic confinement and target gain in laser fusion, but companies also use it for other ratios, and 'Q > 1' is often stated without saying which. This site never publishes a bare Q. Every gain figure is filed as one of six named kinds, each with its numerator and denominator stated: target gain, capsule gain, plasma gain, equivalent gain, engineering gain or wall-plug gain. The tokamak safety factor, lowercase q, is a different quantity.; concept:q-factor-fusion
Reactivity: A measure of how far a reactor's chain reaction is from exactly holding steady: how much the effective multiplication factor, k, differs from 1. Positive reactivity means power is rising, negative means it is falling. It is quoted in pcm (per cent mille, a thousandth of a percent) or in dollars, where one dollar equals the delayed neutron fraction: a reactor given a dollar of reactivity or more is prompt critical. Reactivity coefficients describe how reactivity responds automatically to conditions, such as fuel or moderator temperature or coolant voids, and a coefficient that responds by pushing power back down is what a claim of 'negative feedback' or a 'self-regulating' reactor rests on.; concept:reactivity
Reactivity coefficients (Doppler, moderator temperature and void): The feedbacks that describe how a reactor's reactivity changes on its own as conditions change, without any operator action. The Doppler, or fuel temperature, coefficient: hotter fuel absorbs more neutrons in uranium-238 without fissioning, which pushes reactivity down almost instantly; it is negative in fuel with a large share of uranium-238. The moderator temperature coefficient: in a water-moderated reactor, hotter water is less dense and slows neutrons less well, which usually lowers reactivity. The void coefficient: what happens to reactivity if coolant boils or is lost. It is negative in light-water reactors, where losing water removes moderator. It was large and positive in the RBMK reactors at Chernobyl, where boiling raised power, which caused more boiling, and that contributed to the 1986 accident. CANDU heavy-water reactors have a small positive void coefficient that their control and shutdown systems are designed to handle, and large sodium-cooled fast reactor cores can have a positive sodium void effect, which designers reduce with core shape and rely on other, negative feedbacks such as fuel and structural expansion to outweigh. None of these signs is automatic: they depend on the specific design and on where the core is in its fuel cycle, which is why a passive safety claim should come with the actual numbers.; concept:reactivity-coefficients
Reprocessing: Chemically separating usable material, chiefly unused uranium and plutonium, out of spent nuclear fuel so it can be made into new fuel rather than stored or disposed of as waste. Mixed oxide (MOX) fuel is the main commercial product of reprocessing: it combines plutonium recovered this way with depleted uranium. Reprocessing reduces how much fresh enrichment a fuel cycle needs but raises separate proliferation and cost concerns, which is why only a small number of countries operate reprocessing at commercial scale.; concept:reprocessing
Retracted (status): The status given to a figure that has been formally withdrawn, whether by the source that published it or because a subsequent review found it could not be supported. The record is kept rather than deleted, with its retraction and the reason recorded, because a directory that quietly removes withdrawn claims would lose exactly the information a reader most needs when checking a number against its history.; concept:status-retracted
Roadmap: A company's own statement of what it intends to build or achieve by a future date. A roadmap is a claim about the future, never a result, and this site records it as a planned figure or milestone attributed to the company, kept off every achievement leaderboard regardless of how confidently it is stated. When a roadmap date passes without the promised result, the record stays rather than being quietly removed, so a reader can see which promises were kept.; concept:roadmap
Safety factor (q) q: In a tokamak, the number of times a magnetic field line travels around the machine the long way (toroidally) for each time it goes around the short way (poloidally): a measure of how tightly the field is twisted. It changes from the centre of the plasma to the edge. It controls which instabilities the plasma is vulnerable to: if q at the edge falls to about 2, the plasma almost always disrupts, which limits how much current a tokamak of a given size and field can carry. This lowercase q is unrelated to fusion gain, which is written as capital Q; the letters are shared by coincidence.; concept:safety-factor
Scientific breakeven: The point where the fusion energy released exceeds the energy delivered to the fuel: a target gain above 1 for laser-driven fusion, or a plasma gain above 1 for magnetic confinement. Reached for laser fusion by the National Ignition Facility on 5 December 2022. Not yet reached for magnetic confinement: the best measured result with deuterium-tritium fuel is a plasma gain of 0.63 from JET on 31 October 1997, often quoted as 0.67 because a different figure for the heating power is used as the denominator. Higher figures quoted for other machines, such as JT-60U's, are equivalent gains: calculations for plasmas that contained no tritium. Distinct from engineering breakeven, which counts the whole plant's electricity rather than the energy delivered to the fuel.; concept:scientific-breakeven
Separative work (SWU) SWU: The measure of the effort an enrichment plant spends splitting uranium into an enriched stream and a depleted one, counted in separative work units (SWU). The work rises steeply with the target enrichment. With natural uranium as feed and depleted tails at 0.25% uranium-235, a kilogram of typical 4.5% power reactor fuel needs about 7 SWU and about 9 kg of natural uranium, while a kilogram at 19.75% needs about 41 SWU and about 42 kg. Those are calculated from the standard formula, not measured, but they show why a fleet running on HALEU needs far more enrichment capacity for each kilogram of fuel than today's fleet does. Modern centrifuges use about 40 to 50 kWh of electricity per SWU.; concept:separative-work
Shot: One run of a fusion experiment, from start to finish: a single laser firing on an inertial fusion target, one discharge of a pinch, or one plasma in a tokamak or stellarator. Laboratories differ in the word they use: NIF and JT-60U number shots, JET numbered pulses, and the two mean the same thing. Results in this dataset carry the shot or pulse number where the source gives one, so a figure can be traced to the exact run that produced it. A result quoted with no shot number or date is weaker evidence.; concept:shot
Source independence: A label recording whether a source could have come from the subject itself: independent (could not have come from the subject, such as a peer-reviewed paper or a regulator's docket), operator (the subject's own website or press release), aggregator (a directory that may be copying someone else's material), circular (shown to reproduce this site's own material, never cited as evidence), or unknown, used rather than guessing. A company describing its own machine is always operator, however authoritative it sounds; operator evidence shows what a company says, not that what it says is true.; concept:source-independence
Spent fuel: Nuclear fuel removed from a reactor once it can no longer usefully sustain the chain reaction. It is highly radioactive and still produces decay heat, so it is first cooled for several years in water pools, then usually moved to dry storage casks, and eventually either reprocessed or placed in a deep geological repository; no country yet runs such a repository routinely for commercial spent fuel, though Finland is furthest along. Every fission reactor produces spent fuel, or in a molten-salt reactor a used fuel salt. Designs differ in how much they produce for each unit of energy, in its form, and in how long its radioactivity lasts, so a claim of 'less waste' should say whether it means volume, mass or radioactivity. Back-end capacity is tracked in this dataset as its own category.; concept:spent-fuel
Standard design approval vs design certification: Two different ways the US Nuclear Regulatory Commission can approve a reactor design without approving a specific plant. A design certification is a formal rule the NRC publishes, such as NuScale's, valid for 40 years, and can be referenced directly in a later licence application. A standard design approval is a lighter step that also lets a design be cited in a future application but carries less regulatory weight and is valid for 15 years without renewal. Both leave every other licensing step, including a construction permit and an operating licence, still to come, so 'NRC approved' can describe either one, or neither, depending on which was actually granted.; concept:standard-design-approval
Superconducting magnet: A magnet whose coils carry current with no electrical resistance once cooled far below room temperature, so it can hold a strong field for long periods without the heat and power bill of copper coils. Most large fusion magnets use low-temperature superconductors such as niobium-tin, cooled with liquid helium to about 4 kelvin, as ITER does. Newer designs such as Commonwealth Fusion Systems' SPARC use high-temperature superconducting (HTS) tape, which carries more current in stronger fields and allows a more compact machine. Field strength matters because, at the same beta, a tokamak's fusion power per cubic metre rises with the fourth power of the field.; concept:superconducting-magnet
Superseded (status): The status given to a figure once a later measurement from the same machine beats it. Both figures stay in the record; the superseded one is not deleted, and a device's own page shows its full progression of results, oldest first, so a reader can see how a record improved over time rather than only the current best value.; concept:status-superseded
Topical report: A technical report a developer submits to the US Nuclear Regulatory Commission on one specific subject, such as a fuel's performance or a method of safety analysis, for the NRC to review on its own, before or alongside a licence application. NRC approval of a topical report means the agency accepts that one piece of analysis so it can be referred to later. It is not approval of a design or a plant, and a headline calling it 'NRC approval' describes much less than the phrase suggests.; concept:topical-report
Toroidal and poloidal: The two directions around a ring-shaped (toroidal) device. Toroidal means the long way round, following the ring; poloidal means the short way, around the plasma's cross-section. A tokamak's coils make a strong toroidal field, and the current in the plasma adds a weaker poloidal field; together they twist the field lines into helices, which is what keeps the plasma confined. A stellarator makes the same twist with the shape of its coils instead of a plasma current.; concept:toroidal-poloidal
Tritium breeding: The process by which a deuterium-tritium fusion plant would make its own tritium fuel, since tritium barely exists in nature and a power plant would burn it far faster than outside supply could provide. The standard design is a breeding blanket: a layer containing lithium surrounding the plasma, in which neutrons from the fusion reactions strike lithium nuclei and produce tritium. Success is measured by the tritium breeding ratio, tritium made divided by tritium burned, which must exceed 1 with margin for losses. No fusion device has yet run with a breeding blanket, so none has bred its own fuel. Published breeding ratios are almost always modelled; the few measurements come from small mock-ups placed near neutron sources, which measure tritium made in a sample, not a whole-plant ratio. ITER plans to test breeding blanket modules in its plasma.; concept:tritium-breeding
Vendor design review: An optional, pre-licensing assessment offered by Canada's nuclear regulator, the Canadian Nuclear Safety Commission, in which CNSC staff review a reactor vendor's design against the regulator's requirements before any licence application is filed. The CNSC states explicitly that a vendor design review does not issue a licence, does not certify a design, and does not bind the Commission's later licensing decisions, so it functions as early feedback rather than an approval.; concept:vendor-design-review
verifiedAt verifiedAt: The date a source was last opened during a verification run and the record checked against it. It is set only when the source was actually opened in that run and it supported the value; if it is empty, that check has not yet happened. A related field, verifiedFields, lists exactly which fields were checked, since a record can have some fields confirmed while others have not been looked at again.; concept:verified-at
Zero-power reactor: A reactor, sometimes called a critical assembly, built to reach criticality at such low power, typically watts and at most a few kilowatts, that it makes no useful heat, needs little cooling and builds up little radioactivity. It is used to check that a core's physics matches its calculations: how much fuel it takes to go critical, how much the control rods are worth and how reactivity changes with temperature. Reaching criticality in a zero-power reactor is a real milestone, but it produces no electricity and says little about how a power reactor of the same design will handle heat, materials or years of operation.; concept:zero-power-reactor

Metrics 36

Capacity factor no records yet: Energy actually generated as a fraction of what continuous full-power operation would give. The number that separates a plant that runs from a plant that exists.; metric:capacity-factor
Capsule gain no records yet: Fusion energy divided by energy absorbed by the fuel capsule. Always larger than target gain; quoting it as 'the' gain overstates performance.; Total fusion energy yield ÷ Energy absorbed by the capsule; metric:gain-capsule
Core outlet temperature no records yet: Sets which industrial heat applications a design can serve.; metric:core-outlet-temperature
Cumulative funding raised no records yet: DERIVED by summing deals, never hand-entered. This is what makes '$65M+' and '~$8.94M across 3 rounds' impossible to write down.; metric:cumulative-funding
Electron temperature 1 record: Temperature of the plasma's electrons, usually given in keV. It can be higher or lower than the ion temperature in the same plasma, depending on how the plasma is heated: microwave heating warms the electrons first, while neutral-beam heating mostly heats the ions, and several record temperatures came from 'hot-ion' plasmas in which the ions were much hotter than the electrons. The fusion rate depends on the ion temperature, so an electron temperature quoted as "the" plasma temperature answers a different question and cannot be ranked against ion temperatures.; metric:electron-temperature
Energy confinement time (τ_E) 22 records: How long the plasma retains its energy (τ_E): the energy stored in the plasma divided by the rate at which it loses energy, which equals the heating power only when the plasma is steady. This is NOT the length of the discharge. A machine can hold a plasma for 1,000 seconds with a confinement time of a fraction of a second, and conflating the two is one of the most common errors in coverage. Operators also differ in whether they correct for the plasma's stored energy still rising, so two confinement times for similar plasmas can differ by definition alone.; metric:energy-confinement-time
Engineering gain (Q_eng) Q_eng no records yet: Gross electrical output divided by total electrical input, recirculating power included. No fusion device has yet generated electricity, so none has exceeded zero.; Gross electrical output ÷ Total electrical input including all plant recirculating power; metric:gain-engineering
Equivalent D-T gain (extrapolated) 1 record: Gain a D-D or D-³He shot WOULD have reached had it run on D-T. A calculation, never a measurement. Requires conditions.isSimulated = true and is excluded from achieved leaderboards.; Modelled D-T fusion power ÷ External heating power coupled into the plasma; metric:gain-equivalent
Fuel enrichment no records yet: Percentage of fissile isotope in the fuel. Above 5% and up to 20% is HALEU, which most advanced designs need and for which Western supply is currently thin.; metric:fuel-enrichment
Fusion energy yield 2 records: Total fusion energy released in one shot or pulse. A numerator only: on its own it says nothing about the energy required to produce it. See gain.; metric:fusion-energy-yield
Fusion power 1 record: Thermal power released by fusion reactions. A rate, not a total; the total released in one pulse is fusion energy yield.; metric:fusion-power
Fusion triple product (nTτ_E) nTτ 23 records: Density × ion temperature × energy confinement time. The fairest single comparator across confinement approaches.; metric:triple-product
Headcount no records yet: Employees. A rough proxy for scale, and one of the few figures a private company reliably discloses.; metric:headcount
HTS tape critical current no records yet: Meaningless without conditions: record temperature and applied field in conditions.; metric:hts-critical-current
Ion temperature 23 records: Temperature of the plasma's ions, the nuclei that actually fuse, usually given in keV (1 keV is about 11.6 million kelvin). Every value carries a qualifier, peak or sustained: a peak temperature may have lasted milliseconds or less, a sustained one was held for a stated time, and they are different achievements. Conflating them is a listed red flag. Record figures are also usually the value at the hottest point, the centre of the plasma, not an average over its volume.; metric:ion-temperature
Laser energy on target 1 record: Laser energy delivered on target. The denominator of target gain. The energy drawn from the grid to produce it is far larger, which is why wall-plug gain differs from target gain by orders of magnitude.; metric:laser-energy
Lawson parameter (nτ_E) no records yet: Density × energy confinement time (nτ_E). The Lawson criterion in its original form, before temperature was folded in to give the triple product.; metric:lawson-parameter
Levelised cost of energy no records yet: Levelised cost of energy: lifetime cost divided by lifetime output. Highly sensitive to discount rate, capacity factor and build time, so two LCOE figures are rarely comparable unless all three assumptions are stated.; metric:lcoe
Major radius R0 77 records: Distance from the torus axis to the plasma centre. A design parameter of the machine, not a shot result.; metric:major-radius
Minor radius a 76 records: Plasma cross-section radius. Design parameter.; metric:minor-radius
Net electrical capacity 54 records: Net electrical output to the grid, after the plant's own consumption. Distinct from thermal capacity, and typically lower by a factor of about three.; metric:electrical-capacity
Neutron yield per shot no records yet: Neutrons produced per shot. For D-T, roughly 80% of the energy leaves as 14.1 MeV neutrons, so neutron yield is often the most direct measurement of how much fusion actually occurred.; metric:neutron-yield
On-axis magnetic field 73 records: Toroidal field strength at the plasma magnetic axis (B₀). A design parameter of the machine, not something achieved on a shot. Peak field on the coil is higher and is recorded separately.; metric:magnetic-field-onaxis
Overnight capital cost no records yet: Build cost excluding financing, as if the plant appeared overnight. Widely quoted because it is flattering: real projects pay interest for a decade of construction, and that is where nuclear economics is usually decided.; metric:overnight-capital-cost
Patent families no records yet: Distinct inventions patented, counting one family rather than each national filing. Filing counts inflate the same invention many times over.; metric:patent-families
Peak field on coil no records yet: The highest field anywhere on the magnet, which is what the conductor and structure must survive. Always higher than the on-axis field quoted for the plasma, and the two are not interchangeable.; metric:magnetic-field-peak-coil
Plasma current 64 records: Current driven through the plasma (Iₚ). In a tokamak it generates the poloidal field confinement depends on; a stellarator produces that field from its coils and needs none.; metric:plasma-current
Plasma density 19 records: Number of particles per cubic metre. Fusion rate scales with the square of density, which is why it appears in both the Lawson parameter and the triple product.; metric:plasma-density
Plasma duration 45 records: Wall-clock length of a sustained discharge. Distinct from confinement time, and routinely confused with it in press coverage.; metric:plasma-duration
Plasma gain (Q_plasma) Q_plasma 4 records: Fusion power divided by external heating power coupled into the plasma. The magnetic-confinement counterpart of target gain.; Fusion power produced ÷ External heating power coupled into the plasma; metric:gain-plasma
Refuelling interval no records yet: How long the reactor runs before refuelling. A headline advantage of many small designs, and a claim that only becomes a fact once a reactor has run that long.; metric:refuelling-interval
Repetition rate no records yet: Shots per second. The metric that separates a physics demonstration from a power plant.; metric:repetition-rate
Target gain (Q_target) Q_target 2 records: Fusion energy released divided by laser energy delivered to the target assembly.; Total fusion energy yield of the implosion ÷ Laser energy delivered to the target, including hohlraum losses; metric:gain-target
Thermal capacity 192 records: Heat produced by the reactor core. Roughly three times the electrical capacity for a steam-cycle plant; quoting thermal figures where a reader expects electrical inflates the number without stating an untruth.; metric:thermal-capacity
Tritium breeding ratio no records yet: Must exceed 1 for a D-T plant to be self-sufficient. Almost always modelled rather than measured, so check conditions.isSimulated.; metric:tritium-breeding-ratio
Wall-plug gain 1 record: Fusion energy released divided by total facility energy drawn from the grid for the shot. The number that is almost never quoted, and typically two to three orders of magnitude below target gain.; Total fusion energy yield ÷ Total facility grid energy drawn for the shot; metric:gain-wallplug

Technology approaches 36

Beam-target 2 records: A beam of fast ions fired into a fuel target that is not itself a hot plasma, such as a gas cell or a solid loaded with deuterium or tritium. As each beam ion ploughs through the target it loses its energy to the target's electrons far faster than it is likely to fuse, so only a tiny fraction of the ions fuse before they stop. The fusion energy returned is a small fraction of the energy spent accelerating the beam, so the approach cannot reach net energy gain however the beam or target is arranged. It is commercially useful anyway, as a compact and reliable source of neutrons for imaging and medical isotope production.; approach:beam-target
Boiling water reactor BWR 7 records: Light water boils directly in the core and the steam drives the turbine: one loop instead of two, simpler but with an activated steam circuit. Water is again both coolant and moderator, and fuel is low-enriched uranium oxide, as in a PWR.; approach:bwr
Electrostatic confinement 1 record: Confinement by electric rather than magnetic fields.; approach:electrostatic
Field-reversed configuration FRC 12 records: A self-organised plasma ring that generates its own confining field, with no central coil. Compact and high-beta, but stability over long timescales is the open question.; approach:frc
Fission 17 records: Energy released by splitting heavy nuclei. Every operating nuclear power station works this way; every fusion device in this dataset does not.; approach:fission
Fluoride salt-cooled high-temperature reactor FHR 3 records: Molten salt COOLANT with solid fuel, typically TRISO, and a separate solid moderator, usually graphite. Kairos Power's KP-FHR is the reference example.; approach:fhr
Fusion 2 records: Umbrella term used only where a record’s specific confinement approach is not established by evidence. A narrower term is always preferred when the sources support one.; approach:fusion
Gas-cooled 1 record: Reactors cooled by gas, usually helium or carbon dioxide, allowing higher outlet temperatures than water.; approach:gas-cooled
Heat-pipe microreactor 4 records: Heat moved by sealed heat pipes rather than pumped coolant or forced convection. No primary pumps and no coolant loop, which suits very small transportable reactors. Heat-pipe describes how heat is moved, not the neutron spectrum. Moderator and fuel vary by design: some designs are fast and unmoderated, others use a graphite or beryllium oxide moderator, with TRISO or metal fuel, usually HALEU.; approach:heat-pipe
Heavy-ion ICF no records yet: Heavy ion beams rather than lasers deliver the compression energy. Accelerators are efficient and repeat readily, which lasers of this scale do not. No facility has been built at fusion-relevant scale; the main open problem is focusing multiple intense ion beams onto a millimetre-scale target simultaneously.; approach:icf-heavy-ion
High-temperature gas-cooled reactor HTGR 16 records: Helium-cooled and graphite-moderated, running far hotter than a water reactor. Fuel is TRISO particles, typically formed into pebbles or prismatic blocks, at LEU or HALEU enrichment depending on the design. The high outlet temperature suits industrial process heat as well as electricity.; approach:htgr
Inertial confinement ICF 10 records: Compressing fuel so fast that its own inertia holds it together long enough to burn: nanoseconds, not seconds. Every ICF approach shares the same open problem beyond ignition itself: repeating a shot several times a second at low enough cost per shot to run a power plant, which no driver has yet demonstrated.; approach:inertial-confinement
Inertial electrostatic confinement 3 records: Electric fields rather than magnetic ones confine the ions. Simple to build and well proven as a neutron source; net energy is a different matter.; approach:iec
Laser ICF, direct drive 8 records: Laser beams strike the fuel capsule directly. More efficient than indirect drive in principle, but far more sensitive to how evenly the beams are applied.; approach:icf-laser-direct
Laser ICF, indirect drive 4 records: Laser energy is converted to X-rays inside a small gold or uranium can (a hohlraum), and the X-rays drive the capsule. The NIF configuration. Only about a tenth of the laser energy ends up absorbed by the capsule, most of the rest heating the hohlraum wall, which is the main efficiency cost indirect drive carries relative to direct drive.; approach:icf-laser-indirect
Lead fast reactor LFR 6 records: Molten lead or lead-bismuth coolant, fast and unmoderated. Does not burn in air or react violently with water, unlike sodium, at the cost of weight and corrosion engineering. Fuel is typically nitride or mixed-oxide.; approach:lfr
Levitated dipole 2 records: Confinement around a single magnetically levitated superconducting ring, modelled on how planetary magnetospheres hold plasma. No levitated-dipole device has run at reactor-relevant scale; the main open problem is engineering a superconducting ring that can be levitated, cooled and shielded from neutrons without any physical support.; approach:levitated-dipole
Liquid-metal cooled 2 records: Reactors cooled by molten metal at near-atmospheric pressure. Most designs use no moderator, so the neutrons stay fast. A few do slow their neutrons: Aalo Atomics describes its sodium-cooled Aalo-1 as a thermal reactor, and the IAEA lists Idaho National Laboratory's MARVEL microreactor as liquid-metal-cooled with a thermal spectrum.; approach:liquid-metal
Magnetic confinement MCF 7 records: Holding a hot plasma away from material walls using magnetic fields, sustained for seconds to steady-state rather than the nanosecond pulses of inertial confinement. The open problem shared by every configuration below is exhausting the plasma's heat and helium ash without eroding the wall that must survive it for years, not seconds.; approach:magnetic-confinement
Magnetic mirror 13 records: Plasma trapped between two regions of strong magnetic field that reflect particles back. Simple and linear, but historically leaky at the ends.; approach:mirror
Magnetized target 4 records: A magnetised plasma target compressed mechanically or by an imploding liner. Sits between magnetic and inertial confinement in density and timescale. The main open problem is compressing the liner onto the plasma symmetrically and fast enough to reach fusion conditions before the plasma cools or the liner breaks up.; approach:magnetized-target
Magneto-inertial 3 records: Compression of a magnetised plasma; sits between MCF and ICF in density and timescale.; approach:magneto-inertial
Molten salt reactor MSR 14 records: Fuel DISSOLVED in the salt, which in most designs is fuel carrier and coolant at once. Distinct from a salt-cooled reactor with solid fuel; the two are often confused. Moderator, where one is used at all, is a separate design choice from the salt: thermal-spectrum MSRs are typically graphite-moderated, fast-spectrum MSRs carry no moderator.; approach:molten-salt
Muon-catalysed fusion µCF 1 record: A negative muon temporarily replaces an electron in a D-T (or D-D) molecule, pulling the nuclei roughly 200 times closer together than an electron would, so fusion happens at room temperature without confining a hot plasma at all. Added as its own branch of the fusion tree rather than filed under magnetic or inertial confinement, because it confines nothing: the muon does the work electromagnetically at the molecular scale. No net-energy device has been built: a muon typically catalyses only 100-150 fusions before it is captured by a helium nucleus and lost, and the main open problem is extending that cycle count, or cutting the cost of producing muons, enough to get more energy out than went into making them.; approach:muon-catalysed
Pressurised heavy-water reactor PHWR 7 records: Heavy water (deuterium oxide) kept under pressure as both coolant and moderator, which absorbs far fewer neutrons than ordinary water and so lets the reactor run on natural, unenriched uranium. That is the CANDU design's defining trade: heavy water is expensive to produce, but the fuel needs no enrichment plant. Fuel is natural or slightly enriched uranium oxide, usually in horizontal pressure tubes, refuelled online without shutting the reactor down.; approach:phwr
Pressurised water reactor PWR 34 records: Ordinary (light) water kept under high pressure so it heats without boiling, transferring heat to a separate steam loop; the water is both coolant and moderator. Fuel is low-enriched uranium oxide pellets in zirconium-alloy rods. The most common reactor type in service worldwide.; approach:pwr
Projectile-driven 2 records: A physical projectile accelerated to extreme velocity compresses the fuel on impact. The main open problem is achieving symmetric compression from one, or a few, points of impact, where laser and heavy-ion approaches can spread the drive over many beams.; approach:projectile-driven
Pulsed magnetic compression 1 record: Repeated compression of a plasma by pulsed magnetic fields, recovering energy directly from the expansion rather than through a steam cycle. The main open problem is hardware lifetime: the compression coil and switching electronics must survive millions of repeated pulses for a power plant to be economic.; approach:pulsed-magnetic-compression
Reversed-field pinch 5 records: A toroidal configuration where the field reverses direction at the plasma edge, allowing confinement at much lower external field than a tokamak. That has come at the cost of confinement quality: RFPs have not matched tokamak performance, and closing that gap without the stronger field is the configuration's main open problem.; approach:reversed-field-pinch
Salt-based no records yet: Reactors using molten salt as fuel carrier, coolant, or both.; approach:salt-based
Sodium fast reactor SFR 17 records: Liquid sodium coolant with a fast, unmoderated neutron spectrum. Sodium carries heat superbly at atmospheric pressure, but burns in air and reacts violently with water. Large sodium-cooled cores can have a positive sodium void effect that the core design must offset. Fuel is typically metal alloy or mixed-oxide.; approach:sfr
Spherical tokamak 34 records: A tokamak squeezed towards a cored-apple shape. The tighter aspect ratio holds plasma pressure with less magnetic field, which is efficient, but leaves very little room at the centre for the central column.; approach:spherical-tokamak
Stellarator 25 records: Confinement produced by external coils twisted into complex three-dimensional shapes, so no net current has to be driven through the plasma. That removes the disruptions a tokamak's plasma current can cause and allows continuous operation, at the price of coils that are far harder to design and build.; approach:stellarator
Tokamak 67 records: Doughnut-shaped magnetic confinement using a strong toroidal field plus a current driven through the plasma itself. The most-studied configuration, and the one ITER uses. Its main open problem is heat exhaust at the divertor, where the plasma edge concentrates power onto a small area, together with the risk of disruptions: sudden, uncontrolled losses of the plasma current.; approach:tokamak
Water-cooled 119 records: Reactors cooled by ordinary or heavy water, which in most designs is also the neutron moderator.; approach:water-cooled
Z-pinch 9 records: A current driven straight through a plasma column, whose own magnetic field squeezes it inward. Long set aside as unstable; sheared-flow stabilisation, studied at the University of Washington and pursued commercially by Zap Energy, renewed work on it.; approach:z-pinch

Categories 22

Breeding blanket no records yet: The breeding blanket, which absorbs neutrons to make tritium and capture heat. No commercial-scale blanket has ever operated, and without one D-T fusion has no fuel supply.; cat:blanket
Cryogenics no records yet: Cooling to superconducting temperatures, and the plant that sustains it. Superconducting magnets are only an advantage if the cryoplant works.; cat:cryogenics
Diagnostics no records yet: Instruments that measure what the plasma is doing. Every number in this benchmark exists because a diagnostic produced it.; cat:diagnostics
Divertor no records yet: The component that exhausts heat and helium ash from the plasma edge. Steady-state heat fluxes there rival a rocket nozzle.; cat:divertor
Enrichment no records yet: Raising the fissile fraction of uranium. The chokepoint of the fission fuel cycle, and the step with the heaviest non-proliferation constraints.; cat:enrichment
Gyrotron no records yet: High-power microwave sources for electron-cyclotron heating. A specialist supply chain with very few vendors worldwide.; cat:gyrotron
Heating and current drive no records yet: Systems that heat the plasma and drive current in it: microwaves, neutral beams, radiofrequency.; cat:heating-current-drive
HTS tape no records yet: REBCO and related high-temperature superconducting tape. The binding supply constraint for high-field tokamak programmes.; cat:hts-tape
ICF targets no records yet: Fuel capsules for inertial confinement. A power plant would need several a second; today's research facilities use them at most in the low thousands a year.; cat:targets
Lasers and optics no records yet: Laser drivers and optics. For inertial confinement the laser IS the machine, and its wall-plug efficiency sets the ceiling on any plant.; cat:lasers
Magnets no records yet: Magnet systems that confine the plasma. The dominant cost and engineering risk in most magnetic-confinement designs.; cat:magnets
Neutral beam injector no records yet: Injectors that heat plasma by firing energetic neutral atoms into it, which cross the magnetic field because they carry no charge.; cat:neutral-beam
Plasma-facing components no records yet: Components that face the plasma directly and take the heat and particle flux. This is the hardest materials problem in fusion.; cat:plasma-facing
Power conversion no records yet: Turning heat into electricity: turbines, generators, heat exchangers, steam or gas cycles.; cat:power-conversion
Pulsed power no records yet: Capacitor banks and switching that deliver enormous energy in microseconds, used by z-pinch and inertial approaches.; cat:pulsed-power
Remote handling and robotics no records yet: Remote handling and maintenance. Once a machine is activated, everything inside it must be serviced without a human present.; cat:robotics
Simulation and modelling no records yet: Modelling of plasmas, neutronics, materials and whole plants. Cheaper than experiments, and never a substitute for one.; cat:simulation
Structural and irradiation-tolerant materials no records yet: Structures that survive 14 MeV neutron damage without unacceptable swelling, embrittlement or long-lived activation.; cat:structural-materials
TRISO fuel no records yet: Tri-structural isotropic fuel: fissile kernels in ceramic coatings that retain fission products at high temperature. The safety case for several advanced designs rests on it.; cat:triso-fuel
Tritium handling no records yet: Storage, recovery and accounting for tritium. It is radioactive, mobile, expensive, and the world civil inventory is small enough to constrain the whole sector.; cat:tritium-handling
Used fuel and waste management no records yet: Spent fuel storage, reprocessing and waste management. This is the part of the fuel cycle no design escapes.; cat:used-fuel
Vacuum systems no records yet: Vacuum vessels, pumps and seals. A fusion device is among the largest high-vacuum systems ever built.; cat:vacuum

Regulatory stages 15

Accepted for review 6 records: The regulator has accepted an application for review, a step often called "docketing". It means the paperwork is complete enough to examine, and says nothing about whether it will be approved.; regstage:application-accepted
Application denied 1 record: Terminal for that application. Kept visible, because denials are as informative as approvals and are routinely omitted from company timelines.; regstage:application-denied
Application submitted 6 records: An application has been filed with the regulator. Filing is not acceptance, review, or approval, and applications are regularly withdrawn.; regstage:application-submitted
Application withdrawn no records yet: The applicant withdrew before a decision. Recorded as a first-class outcome, because withdrawals and denials disappear from most trackers, which biases every remaining figure.; regstage:application-withdrawn
Commercial operation no records yet: Declared commercially operational, after commissioning tests are complete. The end of this site's licensing sequence and the start of the economic one, although the regulator's oversight continues for the life of the plant.; regstage:commercial-operation
Construction permit issued 4 records: The regulator has authorised construction at a named site. The strongest pre-construction stage, and distinct from design approval, which is site-independent. It does not allow fuel to be loaded or the reactor to operate; that needs a later licence. Under the US combined-licence route, one licence covers both construction and operation.; regstage:construction-permit-issued
Construction start 2 records: Physical construction of the nuclear plant has begun, usually counted from the first pour of safety-related concrete for the reactor building's foundation. Site clearing and non-nuclear buildings often start earlier and are sometimes announced as the start of construction. Distinct from holding a permit that allows it.; regstage:construction-start
Design approval or certification 2 records: The regulator has approved or certified a design. This is NOT permission to build: a site licence or construction permit is a separate and additional step. Coverage describing design approval as "approved to build" is the most common regulatory error in this sector.; regstage:design-approval
First criticality 1 record: The reactor first sustained a chain reaction, normally at very low power for physics tests. It is the first proof that the built core behaves as designed, not the start of power production: raising power, grid connection and commercial operation follow, often months later. Reaching criticality is the normal way a reactor starts, not an accident.; regstage:first-criticality
Fuel load no records yet: Fuel loaded into the reactor for the first time, the step before first criticality. Fresh uranium fuel is only mildly radioactive; the core becomes highly radioactive once the chain reaction starts and fission products build up.; regstage:fuel-load
Grid connection no records yet: First delivery of electricity to the grid. Frequently reported as "operational", though full commercial operation usually follows months later.; regstage:grid-connection
Operating licence issued 1 record: Permission to operate granted. The point at which a built plant may load fuel and run, sometimes in stages, such as low-power testing before full power is allowed.; regstage:operating-licence-issued
Operating licence submitted no records yet: An application to operate has been filed, typically after construction is under way.; regstage:operating-licence-submitted
Pre-application engagement no records yet: Talking to the regulator before filing anything. No application exists. Coverage describing this as engagement with the regulator is accurate; describing it as progress toward approval is not.; regstage:pre-application
Safety review complete no records yet: The regulator has finished its technical safety evaluation. A major milestone, and still not permission to build.; regstage:safety-review-complete

Relationship types 14

Acquired no records yet: Bought another organisation.; reltype:acquired
Collaborates with no records yet: Works jointly on research or development, typically without a commercial contract.; reltype:collaborates-with
Customer of no records yet: Buys from another organisation.; reltype:customer-of
Employs no records yet: Employs a named person.; reltype:employs
Licensee of no records yet: Holds a licence to another organisation’s technology.; reltype:licensee-of
Member of no records yet: Belongs to a consortium, association or programme.; reltype:member-of
Offtaker of no records yet: Has contracted to buy future output.; reltype:offtaker-of
Operates no records yet: Runs a facility, which may have been built or owned by someone else.; reltype:operates
Owns no records yet: Holds ownership of a facility or asset.; reltype:owns
Partner with no records yet: A stated collaboration between organisations, symmetric by definition.; reltype:partner-with
Regulates no records yet: Holds regulatory authority over another organisation.; reltype:regulates
Spun out of no records yet: Was founded out of a university, laboratory or company. Common in this sector and frequently the origin of the technology.; reltype:spun-out-of
Subsidiary of no records yet: Is majority-owned and controlled by another organisation.; reltype:subsidiary-of
Supplies no records yet: Provides components, materials or services to another organisation.; reltype:supplies

Roles 14

Consortium 10 records: A multi-party body pooling funding or facilities across organisations or states. ITER is the largest example.; role:consortium
Fusion developer 60 records: Develops a fusion device or power plant concept. Carries no implication that the device exists or works.; role:fusion-developer
Government programme 35 records: A state programme rather than a company: the funder, operator or coordinator of a national effort.; role:government-programme
Investor 7 records: Provides capital. Held as a role on the organisation rather than in a separate collection, so one record can be both investor and operator.; role:investor
National laboratory 29 records: A state-funded laboratory operating at national scale, usually running facilities no single university could support.; role:national-lab
Offtaker 2 records: Contracted buyer of future output. This is the role Microsoft occupies with Helion.; role:offtaker
Operator 196 records: Runs a facility, which may have been built by someone else.; role:operator
Reactor developer 64 records: Develops a fission reactor design. Carries no implication that anything is licensed, funded or built.; role:reactor-developer
Regulator 5 records: A government or international body with legal authority to license, inspect or shut down a facility. Regulators are recorded as Organizations, like any other entity in this dataset, so their licensing actions and dockets resolve to a citable page rather than a bare name.; role:regulator
Research institute 134 records: A public or non-profit research body. Frequently operates devices it did not commercially develop.; role:research-institute
Supplier 86 records: Supplies components, materials or fuel to the sector without developing a reactor or device of its own.; role:supplier
Unclassified 17 records: Discovered and captured, but not yet classified. Carries no claim about what the organization does. The autonomous discovery loop lands stubs with this role rather than guessing a real one: a wrong role is a false assertion, while an explicit 'we have not looked yet' is true. Records holding it are excluded from every sector listing and leaderboard until reclassified.; role:unclassified
University 85 records: A degree-granting institution. Many of the smaller devices in this dataset are university machines.; role:university
Utility 49 records: Generates or distributes electricity. Buys, operates or offtakes from plants rather than designing them.; role:utility

Funding instruments 12

Acquisition no records yet: Purchase of one organisation by another.; instr:acquisition
Convertible note or SAFE 1 record: A loan that may convert to equity later. Neither purely debt nor equity until it converts.; instr:convertible
Debt no records yet: Borrowed money, repayable with interest.; instr:debt
Equity round 1 record: Capital raised in exchange for ownership, diluting existing holders. The instrument most often reported as "raised".; instr:equity-round
Grant 1 record: Non-dilutive money that need not be repaid, usually public. Adding grants to equity and calling the sum "total raised" is the most common error in fusion funding figures.; instr:grant
In-kind contribution no records yet: How most ITER-party contributions are actually made.; instr:in-kind
IPO 1 record: First sale of shares to the public.; instr:ipo
Loan guarantee no records yet: A government promise to cover a lender if the borrower defaults. Not cash received, and counting it as funding raised overstates a balance sheet substantially.; instr:loan-guarantee
Milestone-based award 1 record: Paid against delivered technical milestones. The DOE fusion programme's mechanism, and structurally unlike a VC round.; instr:milestone-award
Offtake or power purchase agreement 1 record: Commits capacity, not cash. Recording it as a funding round would overstate capital raised.; instr:offtake-ppa
Prize no records yet: An award won competitively, usually for a demonstrated result.; instr:prize
SPAC merger 1 record: Listing by merger with a shell company already on an exchange. Faster than an IPO and historically associated with more optimistic projections.; instr:spac

Lifecycle status 12

Acquired 2 records: Bought by another organisation. The record is retained, with the acquirer named, because history does not stop at acquisition.; status:acquired
Active 474 records: The organisation is trading or operating. This says nothing about whether its technology works: an active company may have built nothing.; status:active
Cancelled 1 record: Abandoned before completion. A first-class outcome in this dataset, not an absence of one.; status:cancelled
Commissioning 3 records: Built, and being brought into operation. Between construction and operating, and frequently the longest phase.; status:commissioning
Decommissioned 14 records: Permanently shut down and being dismantled.; status:decommissioned
Defunct 1 record: Ceased to exist. The record stays: a directory that quietly deletes failures overstates the sector’s success rate.; status:defunct
Operating 197 records: The facility is running. For a device this means it produces plasma or is critical, not merely that the building stands.; status:operating
Planned 28 records: Announced or designed, but not under construction. Permanently ineligible for achievement leaderboards: a target is not a result.; status:planned
Shut down 26 records: Permanently ceased operation. The record is kept rather than deleted, because the history is part of the evidence.; status:shutdown
Stealth no records yet: Operating without public disclosure of what it is doing. Recorded when existence is established but activity is not.; status:stealth
Suspended 11 records: Not operating, but not permanently closed: the operator or regulator lists it as temporarily or indefinitely shut down, for refurbishment, fuel supply, licensing or funding. It may restart. The IAEA Research Reactor Database calls this 'temporary shutdown' or 'extended shutdown'.; status:suspended
Under construction 25 records: Physical construction has begun on site. Distinct from a construction permit, which is permission to begin and often precedes the first concrete by years.; status:construction

Product types 11

Component no records yet: A discrete part supplied into someone else’s system: a magnet, a valve, a vacuum vessel.; ptype:component
Energy offtake no records yet: A contract to buy future output. The buyer is taking a position on a plant existing, which is why offtakes are tracked as deals rather than as revenue.; ptype:energy-offtake
Fuel no records yet: Fuel as a product: TRISO pebbles, fuel assemblies, tritium, deuterium.; ptype:fuel
Instrument or diagnostic no records yet: Diagnostic or measurement equipment, from neutron detectors to interferometers.; ptype:instrument
Isotope or radiopharmaceutical no records yet: Near-term revenue for several fusion-adjacent firms, and often the real business model behind a fusion roadmap.; ptype:isotope
Material no records yet: A material supplied as such: lithium, beryllium, tungsten, HTS tape, structural steel.; ptype:material
Power plant design no records yet: A complete plant concept, of which instances may be built. The design is the Product; a plant built from it is a Facility.; ptype:power-plant-design
Reactor module no records yet: A factory-built reactor unit intended to be shipped and installed rather than constructed on site. This is the central premise of the SMR case.; ptype:reactor-module
Service no records yet: Work sold rather than a thing: engineering, testing, irradiation, decommissioning, consultancy.; ptype:service
Software no records yet: Simulation, control or analysis software sold or licensed as a product.; ptype:software
Subsystem no records yet: An integrated assembly performing a function: a heating system, a divertor, a fuel-handling plant.; ptype:subsystem

Fuels 10

Deuterium–deuterium D-D 10 records: Deuterium only. Needs far higher temperature than D-T, but the fuel is abundant in seawater and no tritium breeding is required. It is not neutron-free: about half of D-D reactions release a 2.45 MeV neutron, and the tritium made by the other half can fuse in turn, releasing 14.1 MeV neutrons.; fuel:d-d
Deuterium–helium-3 D-³He 1 record: Deuterium with helium-3. Most of the energy is released as charged particles rather than neutrons, which allows direct electrical conversion, but deuterium also fuses with itself in the same plasma, so some neutrons are still produced. Helium-3 is vanishingly scarce on Earth.; fuel:d-he3
Deuterium–tritium D-T 46 records: Lowest ignition threshold; 80% of energy in 14.1 MeV neutrons. The reference fuel for every near-term net-gain claim.; fuel:d-t
High-assay low-enriched uranium HALEU 21 records: High-assay low-enriched uranium: uranium enriched to between 5% and just under 20% uranium-235, as the US Department of Energy defines it. It is still formally low-enriched, because the international line for highly enriched uranium is 20%, which is why many designs specify 19.75%. The binding supply constraint for many advanced reactor designs. The 5 to 10% part, which carries lighter US physical-security requirements than material from 10% up, is also tagged as LEU+. A developer calling fuel in that band 'LEU' is making a labelling choice, not describing a different enrichment.; fuel:haleu
LEU+ (5–10% U-235) LEU+ 1 record: Uranium enriched to between 5% and 10% uranium-235, the band the World Nuclear Association and the industry call LEU+. It lies inside the HALEU range, but is tracked separately for a concrete reason: US security rules place uranium below 10% enrichment in the lightest category (Category III) whatever the quantity, while 10 kg or more of uranium-235 in uranium enriched to 10% or more falls in Category II, which needs heavier protection. A design that stays under 10% therefore carries materially lighter security requirements. Aalo Atomics's fuel at roughly 8% uranium-235 is the reference case: the company calls it 'LEU' in its own materials, but by the 5 to 20% HALEU definition it is HALEU, in the LEU+ band.; fuel:leu-plus
Low-enriched uranium 17 records: Uranium enriched to 5% uranium-235 or less, typically 3 to 5%: the range almost all power reactors use. Formally, anything below 20% is low-enriched; this site holds the band from 5% to just under 20% separately as HALEU, with its 5 to 10% part also tagged as LEU+. A fuel above 5% uranium-235 belongs in one of those, never here, whatever a developer's own materials call it.; fuel:leu
Mixed oxide fuel MOX 4 records: Mixed oxide fuel: plutonium oxide blended with uranium oxide, usually depleted uranium, with commercial MOX typically 7 to 11% plutonium. The plutonium comes from reprocessed spent fuel, so MOX needs no fresh enrichment but depends on a reprocessing plant. Plutonium gives a smaller delayed neutron fraction than uranium-235, one reason thermal reactors usually load MOX into only part of their core.; fuel:mox
Proton–boron-11 p-¹¹B 6 records: Called aneutronic because the main reaction releases no neutrons, though side reactions and impurities still produce some. Requires at least an order of magnitude more triple product than D-T. A p-¹¹B roadmap that skips D-T is a listed red flag.; fuel:p-b11
Thorium 4 records: Thorium-232, which is fertile rather than fissile: after absorbing a neutron it becomes uranium-233 through two radioactive decays, the second taking about a month, and only then can it fission and sustain a chain reaction. A thorium reactor therefore needs a starting charge of fissile material. Thorium is about three times more abundant than uranium in the Earth's crust.; fuel:thorium
TRISO particle fuel TRISO 10 records: Fuel kernels about the size of a poppy seed, each wrapped in layers of carbon and silicon carbide, so every particle acts as a tiny containment for its own fission products. The US Department of Energy reports that irradiated particles came through more than 300 hours of testing at temperatures up to 1,800 °C with no to minimal damage. That record covers the particle designs, temperatures and burnups actually tested; a claim outside that envelope is not covered by it.; fuel:triso

Funding rounds 9

Bridge no records yet: Short-term financing between rounds. Frequently a signal about timing rather than about progress.; round:bridge
Growth no records yet: Large round raised outside the lettered sequence, often from crossover or strategic investors.; round:growth
Pre-seed no records yet: The earliest institutional money, usually before a working prototype.; round:pre-seed
Seed no records yet: First substantial round, typically funding a first demonstration.; round:seed
Series A no records yet: First priced venture round, usually after a technical result.; round:series-a
Series B no records yet: Scaling round, typically funding a larger machine.; round:series-b
Series C no records yet: Later-stage round, typically toward a pilot plant.; round:series-c
Series D no records yet: Late-stage round.; round:series-d
Series E no records yet: Very late-stage round; in this sector often the one funding first-of-a-kind construction.; round:series-e

Investor types 9

Angel no records yet: An individual investing personally.; invtype:angel
Corporate venture no records yet: A corporate investment arm, typically with strategic as well as financial motives.; invtype:corporate-vc
Crossover or public-market fund no records yet: A public-markets investor buying into private rounds, usually late and often ahead of a listing.; invtype:crossover
Development bank no records yet: A multilateral or national development institution lending against policy goals.; invtype:development-bank
Family office no records yet: Private capital managed for one family. Often the earliest and most patient money in this sector.; invtype:family-office
Government agency no records yet: A public body deploying public money, usually against policy objectives rather than returns.; invtype:government-agency
Sovereign wealth fund no records yet: A state investment fund.; invtype:sovereign-wealth
Strategic corporate no records yet: An operating company investing for commercial advantage: a utility, a supplier, a customer.; invtype:strategic
Venture capital no records yet: A fund investing other people’s money for financial return.; invtype:venture

Facility types 8

Enrichment plant no records yet: A facility that raises the fissile fraction of uranium.; ftype:enrichment-plant
Fuel fabrication plant no records yet: A facility that manufactures reactor fuel.; ftype:fuel-fabrication
Fusion device 133 records: An experimental machine that confines plasma for fusion research: tokamaks, stellarators, mirrors, pinches. Laser and pulsed-power drivers are ftype:laser-facility instead.; ftype:fusion-device
Laser facility 7 records: A driver facility that compresses fuel using laser or pulsed-power energy rather than confining it magnetically.; ftype:laser-facility
Manufacturing site no records yet: A factory producing components, modules or materials for the sector.; ftype:manufacturing
Power reactor 20 records: A reactor built to generate power commercially, rather than to conduct research.; ftype:power-reactor
Research reactor 147 records: A reactor operated for experiments, isotope production or training rather than for grid power.; ftype:research-reactor
Test stand no records yet: A rig for testing components or subsystems outside a complete machine.; ftype:test-stand

Licences 8

CC BY no records yet: Attribution required. Populate the attribution field or the gate fails.; license:cc-by
CC BY-SA no records yet: Creative Commons Attribution-ShareAlike: reusable with credit, but derivatives must carry the same licence. That is why it is tracked separately from CC BY.; license:cc-by-sa
Fair dealing (identification) no records yet: A third party's copyrighted image shown at reduced size to identify what an organisation makes, with attribution and a link to the source. Weaker than trademark-nominative use, which covers logos: a mark may be reproduced to name its owner, whereas a product photograph is a creative work and its use rests on fair dealing or fair use. Recorded distinctly so these can be found and removed on request.; license:fair-dealing-identification
Nominative trademark use no records yet: Using a mark to identify the entity it belongs to, in an editorial directory, unmodified and without implying endorsement. The normal basis for logos here. It is a defensible position, not a licence granted to us: the mark stays the owner's.; license:trademark-nominative
Permission granted no records yet: Explicit written permission. Record who granted it and when in attribution.; license:permission-granted
Press kit no records yet: Published by the owner for media use. Check the accompanying terms, which often restrict modification.; license:press-kit
Public domain no records yet: No copyright restrictions apply. Common for work produced by US federal agencies.; license:public-domain
Unknown no records yet: Placeholder only. An asset with license:unknown must never be published; the gate rejects it. Use it to park a capture pending a licence determination, never to ship one.; license:unknown

Assertion types 7

Diagram or schematic no records yet: A technical drawing or schematic.; atype:diagram
Facility photograph no records yet: A photograph of a plant, device or site.; atype:facility-photo
Headshot no records yet: A photograph of a person.; atype:headshot
Icon no records yet: Square mark without the wordmark, for favicons and dense table rows.; atype:icon
Logo no records yet: The organization's primary identifying mark.; atype:logo
Product image no records yet: A picture of what an organisation makes. A copyrighted work, unlike a logo, so it is shown for identification with attribution.; atype:product-image
Wordmark no records yet: A logo consisting of the organisation’s name set as type, rather than a symbol.; atype:wordmark

Product stages 6

Commercial no records yet: Sold and deployed to customers.; pstage:commercial
Concept no records yet: Described on paper. No hardware exists.; pstage:concept
Discontinued no records yet: No longer developed or offered. Kept in the record rather than deleted.; pstage:discontinued
In development no records yet: Actively engineered, with hardware being built or tested, but not yet a working whole.; pstage:development
Prototype no records yet: A working article exists and is being tested. Not necessarily at full scale.; pstage:prototype
Qualified no records yet: Formally demonstrated against a specification or certified by an authority.; pstage:qualified

Programmes 6

ARPA-E no records yet: US advanced research agency for energy, funding high-risk work including several fusion concepts.; prog:arpa-e
DOE Milestone-Based Fusion Development Program no records yet: US Department of Energy programme paying private fusion developers on delivery of agreed technical milestones rather than up front.; prog:doe-milestone-based-fusion
EUROfusion no records yet: The European consortium coordinating fusion research across member states. It ran JET's scientific programme; UKAEA operated the machine until it closed in December 2023.; prog:eurofusion
ITER no records yet: The international tokamak under construction in France, funded by seven members representing over half the world’s population.; prog:iter
NRC Advanced Reactor Content of Application Project no records yet: The US Nuclear Regulatory Commission’s Advanced Reactor Content of Application Project, defining what an advanced-reactor application must contain.; prog:nrc-arcap
UK STEP no records yet: Spherical Tokamak for Energy Production: the UK programme to build a prototype fusion power plant at West Burton.; prog:uk-step

Milestone types 5

Commercial no records yet: A market event: a contract, an offtake, a first sale.; mstype:commercial
Construction no records yet: A physical build milestone: first concrete, module delivery, vessel installation.; mstype:construction
Financial no records yet: A funding or corporate-finance event.; mstype:financial
Regulatory no records yet: A step in a licensing process, decided by a regulator rather than the developer.; mstype:regulatory
Technical no records yet: An achievement of the machine or the physics.; mstype:technical

Sectors 4

Fission 237 records: Advanced and small modular fission reactor developers, plus established reactor vendors.; sector:fission
Fuel cycle 28 records: Mining, conversion, enrichment, fabrication, reprocessing, waste and tritium supply.; sector:fuel-cycle
Fusion 214 records: Organizations pursuing energy from nuclear fusion, and their dedicated supply chain.; sector:fusion
Supply chain 58 records: Components, materials, instrumentation and services sold into fusion or fission programmes.; sector:supply-chain

Size classes 3

Large reactor no records yet: A fission reactor design rated above 300 MWe electrical output per unit, beyond the IAEA's usual SMR line. The World Nuclear Association's count of 441 operable power reactors with about 404 GWe of capacity (August 2026) gives an average of about 920 MWe per reactor. This class is DERIVED from a design's stated per-unit electrical rating (metric:electrical-capacity), never assigned. A design with no capacity measurement on record has no scale class. The IAEA's own wording is 'typically' up to 300 MW(e) per unit for SMRs and 'typically up to 10 MW(e)' for microreactors, and it does not say net or gross; many developers do not either, so the rating is used as the developer states it.; scale:large
Microreactor no records yet: A fission reactor design rated at up to about 10 MWe electrical output per unit, following the IAEA's description of microreactors as 'typically up to 10 MW(e)'. The US Department of Energy describes microreactors by heat output instead, as 1 to 20 MW thermal. This class is DERIVED from a design's stated per-unit electrical rating (metric:electrical-capacity), never assigned by a steward's judgement of what 'feels' small. A design with no capacity measurement on record has no scale class. The IAEA's own wording is 'typically' up to 300 MW(e) per unit for SMRs and 'typically up to 10 MW(e)' for microreactors, and it does not say net or gross; many developers do not either, so the rating is used as the developer states it.; scale:microreactor
Small modular reactor SMR no records yet: A fission reactor design rated at up to 300 MWe electrical output per unit, following the IAEA's description of SMRs as 'typically up to 300 MW(e) per unit'. This class is DERIVED from a design's stated per-unit electrical rating (metric:electrical-capacity), never assigned. A microreactor (up to about 10 MWe) is the smallest tier within this range in common usage, but is held here as its own class because readers filter for it separately. A design with no capacity measurement on record has no scale class. The IAEA's own wording is 'typically' up to 300 MW(e) per unit for SMRs and 'typically up to 10 MW(e)' for microreactors, and it does not say net or gross; many developers do not either, so the rating is used as the developer states it.; scale:smr