A nuclear reactor splits heavy atoms (uranium-235, plutonium-239) in a controlled chain reaction. Each split releases heat; the heat boils water (or another working fluid) to spin a turbine. Everything that distinguishes one reactor from another — coolant, fuel form, neutron spectrum, pressure, output temperature — is downstream of one engineering choice: how to move that heat out of the core safely and economically.
A small modular reactor (SMR) is any reactor under 300 MWe designed to be factory-built in modules and shipped to site, rather than custom-built in place. The bet is that factory production cuts cost the way it did for cars and aircraft — turning each unit from a one-off megaproject into the 50th, 100th, or 300th instance of a stable design. A microreactor is the same logic taken further: under 20 MWe, truck-portable, and sited where a diesel generator would otherwise sit.
Why this is happening now: hyperscaler data-center load is growing faster than the grid can add firm capacity; coal is retiring; gas turbines have multi-year backlogs; and policy (Inflation Reduction Act, ADVANCE Act, the 2025 DOE Reactor Pilot Program) has converged on nuclear as the only zero-carbon dispatchable baseload that scales. Thirty-four companies in this map are racing the same two-bottleneck: regulatory pathway and HALEU fuel supply.
Every reactor in this market map belongs to one of four families, defined by coolant choice. Coolant choice cascades into fuel form, operating temperature, regulatory pathway, and ultimately whether the design qualifies as Generation III+ or Generation IV.
Nuclear power is 84 years old as a controlled chain reaction, 75 as a source of electricity, 70 as a commercial industry. Every SMR developer in the map descends from one of three foundational programs (1942–1969): the US national-lab reactor experiments, the US Navy's pressurized-water program, or the German pebble-bed gas experiments. The 2020s "renaissance" is not new physics — it is the commercialization of physics proven 50+ years ago and shelved during the post-TMI freeze.
Reactor safety is best understood through three escalating concepts.
Gen IV reactor families are designed around inherent safety wherever possible, falling back to passive where inherent isn't achievable. This is what enables shrinking the Emergency Planning Zone (EPZ) from 10 miles to the site boundary — and dramatically changes siting economics. A reactor with a 0.5-mile EPZ can sit inside an industrial park or behind a data-center meter; a reactor with a 10-mile EPZ cannot.
Reactors are defined as much by their fuel as by their coolant. Six fuel cycles matter in 2026.
The waste-burning thesis — Moltex SSR-W is the cleanest example — is structurally similar to thorium: design a reactor that consumes existing spent fuel as fuel, transforming a stranded liability into a revenue-generating asset. See the next section for the deeper case.
Thorium has been "the next nuclear fuel" for 70 years. The first MSR ever built — the 1965 Molten Salt Reactor Experiment at Oak Ridge — ran on U-233 bred from thorium. The reason it didn't take over was political: the AEC consolidated funding behind sodium fast breeders in 1976, and thorium R&D effectively ended in the West for 40 years.
The 2020s revival is not about uranium scarcity (there's plenty) and not about proliferation resistance (overstated). It is about waste. Thorium reactors can be engineered to consume the long-lived transuranic isotopes that make spent LWR fuel a 100,000-year storage problem, transmuting them into shorter-lived isotopes while producing electricity. The pitch reframes nuclear waste from a stranded liability into a revenue-generating feedstock — which is what's drawing European industrial capital to Thorizon (Netherlands), Stellaria (France, CEA spin-out), and Saltfoss (Denmark, formerly Copenhagen Atomics).
The catch: every one of these designs is FOAK twice over — first-of-a-kind reactor and first-of-a-kind fuel cycle. None of them have operated. The thorium thesis is technically credible and commercially unproven; it is among the highest-risk, highest-asymmetric-upside corners of the entire SMR universe.
Levelized cost of energy ($/MWh) for new-build US generation. Bars show the low–high range for unsubsidized utility-scale projects. Lower is cheaper. The dashed line at $100/MWh is the threshold under which SMRs would be competitive with gas baseload.
The nuclear industry organizes reactor designs into four generations.
Three pathways exist for getting an advanced reactor to first criticality in the US.
Companies using the DOE pathway are betting that (a) the test reactor proof-of-concept will be fast enough to matter, and (b) NRC will accept DOE-validated technology more readily for the follow-on commercial license. Whether this bet pays off is the central regulatory question of the SMR era.
In this market map, the REG column shows each company's current status — OP, CP, UR, PA, or NE. The NE category is dominated by companies pursuing the DOE pathway instead of NRC.
Reference vocabulary, organized by domain. Every dotted-underline term elsewhere in the app links here.