Smaller Reactors, New Fuel Cycles—and Some Very Old Questions
For decades, the phrase “nuclear power plant” called to mind a massive concrete complex whose location everyone knew. Now nuclear energy is being presented in a different form: factory-built small modular reactors, truck-transportable microreactors, molten-salt systems, renewed experiments with thorium and, farther out, fusion. Supporters describe a source of dependable, low-carbon power that can complement wind and solar. Skeptics hear a familiar promise—that the next reactor will finally solve the problems of cost, accidents and waste. Both views contain part of the truth.
A renaissance—before a building boom
The renewed interest is real, but much of it is still a renaissance of policy, investment and contracts rather than completed American reactors. Illinois first opened the door to small modular reactors and then, beginning in 2026, repealed its broader moratorium on new nuclear construction. Governor J.B. Pritzker has directed state agencies to examine possible sites as Illinois confronts projected power shortfalls. Globally, small reactors are operating in Russia and China, but most of the more than 80 designs tracked by the International Atomic Energy Agency remain under development, construction or regulatory review.
The economic claim is straightforward: manufacture repeatable modules in factories, transport them to prepared sites and add capacity as needed, avoiding the delays and cost overruns that burdened recent large plants. That promise has not yet been demonstrated at commercial scale in the United States. Smaller units cost less individually, but may cost more per unit of electricity unless standardization and repeated production deliver the expected savings.
Are data centers driving the revival?
Yes—although they are not the only cause. Electrification, manufacturing, coal-plant retirements, climate goals and energy-security concerns all matter. Data centers have changed the urgency. The International Energy Agency projects global data-center electricity use to more than double between 2024 and 2030. These facilities need enormous amounts of power around the clock, often faster than new transmission can be built. Nuclear power appeals to technology companies because it produces little operational carbon and runs at high capacity through day, night and winter.
The connection is especially visible in Illinois. Meta signed a 20-year agreement supporting continued operation of Constellation's Clinton reactor after state credits expire in 2027. Microsoft is backing the proposed restart of Pennsylvania's undamaged Three Mile Island Unit 1, renamed the Crane Clean Energy Center. Google, Amazon and other firms have also contracted with advanced-reactor developers. These deals do not prove that new reactors will be cheap or arrive on schedule, but they give developers something historically scarce: wealthy customers willing to make long-term commitments.
Small enough to be next door?
An SMR generally produces up to 300 megawatts of electricity per module—small compared with a conventional reactor, but still an industrial power plant. A microreactor may produce only a few megawatts and be shipped by truck or rail, with proposed uses at remote communities, mines, military bases and data centers. “Deployable” means factory-fabricated and transportable; it does not mean a reactor can be parked beside a subdivision like a backup generator.
Any commercial U.S. reactor still requires federal licensing, security, environmental review, an exclusion area and an approved site. However, the Nuclear Regulatory Commission now allows emergency-planning zones for qualifying advanced reactors to be sized according to the demonstrated hazard; for some designs, that zone could end at the site boundary. This makes closer siting more plausible. It also makes transparent local notice, land-use standards, evacuation assumptions, water protection, decommissioning funds and public access to safety information more important—not less.
Can a reactor be truly accident-proof?
No serious energy system is risk-free, and no fission reactor should be described as absolutely safe from every accident, flood, tornado, earthquake, fire, design defect, cyberattack or deliberate attack. New designs can nevertheless reduce particular hazards substantially. Some shut down through gravity and natural circulation rather than pumps and operator action. High-temperature gas reactors use heat-tolerant coated fuel particles. Molten-salt reactors operate at low pressure and may drain fuel into passive cooling tanks. Smaller cores contain less radioactive material, limiting the possible source term.
Each advantage comes with qualifications. Molten salts can corrode structural materials. Sodium-cooled reactors avoid high pressure but use a coolant that reacts with air and water. Novel fuels require new manufacturing and inspection systems. A fleet spread among many sites creates more locations to secure, staff, inspect, refuel and eventually dismantle. Passive safety reduces dependence on machinery and human response; it does not repeal material failure, error or malice.
Past accidents also resist slogans. Chernobyl contaminated large areas and forced lasting displacement. Fukushima caused immense evacuation, cleanup and social costs, although the United Nations scientific review found no documented radiation-attributable health effects among residents and does not expect a detectable future increase. A reactor accident is unlikely to contaminate “the whole planet” at dangerous levels, but Chernobyl proved that contamination can cross borders and impose regional consequences for generations.
Waste: the question that has not gone away
The strongest unresolved argument against fission is not the volume of waste, which is relatively small, but its longevity and the obligation it creates. The United States has more than 90,000 metric tons of commercial spent fuel and still no operating permanent repository; about 2,000 metric tons are added each year. Fuel is first cooled in pools and then may be moved to robust dry casks. Those are credible interim measures, but “interim” has become an inheritance passed from one generation to the next.
Advanced reactors do not erase this responsibility. Some may use fuel more efficiently, consume portions of existing actinides or produce less long-lived transuranic waste. Others create different waste streams requiring new handling and disposal pathways. Every fission system still produces radioactive fission products, activated equipment and material that must be isolated. A responsible expansion should therefore require a funded waste and decommissioning plan before—rather than decades after—the first reactor is built.
Thorium: promising fuel, not a magic reactor
Thorium is abundant and can be converted inside a reactor into fissile uranium-233. It may offer better fuel utilization and less production of some very long-lived actinides. Interest has grown alongside molten-salt technology, and China's experimental two-megawatt-thermal reactor reached full power in 2024 and reported thorium-to-uranium fuel conversion in 2025. That is a significant research milestone, not a commercial power station.
Thorium does not itself eliminate meltdown risk, proliferation concerns or waste. A thorium cycle must be started with fissile material; uranium-233 requires safeguards; chemical processing can be complex; and fission products still require long-term isolation. Its potential is worth serious research, but claims that it makes nuclear power clean, cheap and harmless all at once run well ahead of the evidence.
Fusion: a different promise on a longer horizon
Fusion joins light nuclei rather than splitting heavy ones. It cannot sustain the same runaway chain reaction as fission, uses only small amounts of fuel at a time and produces no spent fuel of the conventional reactor kind. The National Ignition Facility has repeatedly achieved fusion ignition—more energy released from the tiny target than laser energy delivered to it. But the entire facility still consumes far more energy than the fusion shot produces, and no fusion plant has generated commercial electricity.
ITER, the major international magnetic-fusion experiment, now targets deuterium-tritium operations around 2039. Private companies promise faster progress, but formidable tasks remain: durable materials under neutron bombardment, tritium breeding, continuous operation, heat capture, maintenance and competitive cost. Fusion should be pursued vigorously, but it cannot be the plan for power shortages in the 2020s or early 2030s. It will also produce activated reactor materials and tritium-management challenges, though not the same long-lived waste profile as fission.
Clean power beyond the atom
The most promising non-nuclear competitor for round-the-clock clean power may be next-generation geothermal. Enhanced and closed-loop systems apply modern drilling methods to reach heat in far more locations than conventional geothermal plants. The IEA estimates that, with continued cost reductions, geothermal could meet as much as 15 percent of global electricity-demand growth through 2050. It is firm, compact and deployable in modules, although drilling cost, induced seismicity, water use and geological uncertainty require careful management.
Long-duration storage is advancing as well: iron-air and flow batteries, pumped water, compressed air, thermal storage and hydrogen can shift renewable electricity across hours or days. None is a universal substitute for generation, and seasonal storage remains difficult. The near-term answer is therefore likely to be a portfolio—wind and solar, stronger transmission, efficiency, flexible demand, storage, geothermal, existing nuclear and, where the evidence supports it, carefully sited new reactors.
A standard worthy of the promise
The nuclear renaissance deserves neither reflexive rejection nor a blank check. Before communities accept reactors that are smaller, more numerous and potentially closer to where people live, developers should demonstrate their safety claims at full scale; disclose accident and security assumptions; fund independent regulation, waste custody and decommissioning; and give host communities a meaningful voice. A design is not proven because a computer model says it is safe, and a waste problem is not solved because the container is expected to outlive us.
Nuclear power may help build a reliable, low-carbon grid. But the measure of progress is not simply whether we can deploy a reactor. It is whether we can accept responsibility for the entire chain—from fuel mining and enrichment to daily operation, credible worst cases, dismantlement and stewardship long after the electricity has been used.
Questions a host community should ask
Credible worst case. What is the largest release that remains physically possible—not merely the accident the design is expected to prevent—and what independent evidence supports that estimate?
Boundary and emergency planning. Where do the exclusion and emergency-planning zones end? Who decides, what assumptions are used, and how would schools, hospitals, farms, wells and waterways be protected?
Waste custody. Exactly what waste will the reactor create, where will it be stored, who owns it after shutdown and what happens if the operator or vendor no longer exists?
Security and staffing. How many trained operators and security personnel will be present? What protection is required against cyberattack, sabotage, theft and an extended loss of outside power?
Financial assurance. Are decommissioning, waste storage, accident response and site restoration funds protected in advance and periodically adjusted for inflation and new information?
Public accountability. Will inspection findings, abnormal events, radiation monitoring and ownership changes be reported promptly in language residents can understand? A smaller emergency zone should never mean a smaller obligation to disclose risk.