Small modular reactors are factory-built fission plants of roughly 50-350 MWe sold as firm 24/7 power for datacentres, backed by hyperscaler PPAs worth a ~$6.8B market in 2025 but with first deployments not expected before 2030 and bulk delivery in 2031-2035 ref.
Summary
A small modular reactor is a fission plant designed to be built in a factory rather than on site, shipped in modules, and deployed in multiples to reach the required output. The bet is that serial manufacture replaces the bespoke megaproject engineering that has made large nuclear slow and expensive, and that smaller units simplify siting, financing and licensing. Designs differ substantially: NuScale’s 77 MWe light-water module is the only fully design-certified SMR in the US (2023), while X-energy’s Xe-100 is a high-temperature gas-cooled design in an approximately 18-month NRC construction-permit review, and Kairos and Oklo remain in pre-licensing ref.
Three parameters decide the outcome. First, schedule: the only near-term firm nuclear power for a hyperscaler is the Microsoft-Constellation Three Mile Island restart in 2028, which is an existing reactor, not an SMR, and it sets a ~$65/MWh PPA benchmark that new-build SMRs must approach ref. Second, cost: NuScale spent about 15 years reaching regulatory approval and then cancelled its first project for lack of customer interest, and the LCOE case for new nuclear has been the persistent weak point even where the baseload argument is strong ref. Third, fuel: HALEU is the binding constraint, with DOE committing $2.7B over ten years in January 2026 for domestic enrichment, the NDAA directing at least 21 MT to developers by June 2026, and Centrus having delivered 900 kg while building toward 12 MT/yr ref.
Evidence base
- SMR-for-datacentre market assessed at roughly $6.8B in 2025 across signed PPAs, early EPC and government grants; 2030 is the pivotal first-deployment year with bulk deployment 2031-2035 ref.
- NuScale’s 77 MWe design is the only fully design-certified SMR in the US (2023); X-energy’s Xe-100 is in an ~18-month construction-permit review with a first unit at Dow Texas around 2028; Kairos and Oklo remain in pre-licensing ref.
- Google-Kairos covers up to seven SMRs and 500 MW with first unit 2030 and completion 2035; Amazon-X-energy covers four Xe-100s against a claimed 11.5 GW / 144-reactor pipeline in the early 2030s; Oklo has a 500 MW Equinix PPA ref.
- HALEU is the binding fuel constraint: DOE committed $2.7B over ten years in January 2026, the NDAA directed at least 21 MT to developers by June 2026, and Centrus has delivered 900 kg while building 12 MT/yr capacity ref.
- Gas turbine delivery windows have stretched to 5-7 years with the power crunch most severe through 2028-29, before supply begins catching up ref.
- Average interconnection waits of five years are described as the defining constraint on whether new datacentres can come online at all ref.
Open questions
- Does X-energy receive its Xe-100 construction permit and does the Dow Texas first unit hold its ~2028 date, or does it slip into the 2030s with the rest of the fleet?
- What is the delivered PPA price of the first commercial SMR, and how far above the ~$65/MWh Three Mile Island restart benchmark does it land?
- Can HALEU enrichment scale from Centrus’s 900 kg delivery and 12 MT/yr build-out to the tonnage required by an 11.5 GW pipeline, and on what schedule?
- If gas turbine supply normalises after 2029 as forecast, do hyperscalers hold their SMR offtake commitments or renegotiate them?
Assessment drafted 2026-08-31 from up to 15 KB sources using the technology-scorecard framework; scores are a draft read pending review.