Nuclear & SMR Power for AI Data Centers — Strategy
Updated 6/19/2026
Where Nuclear & SMR Power for AI Data Centers is heading over the next 12 months, grounded in product-axis evidence and verbatim demand from the last 90 days. The judgment column is the engine's read — operators verify and refine.
Product trajectories
Xe-100 high-temperature gas reactor — HTGR pebble-bed SMR (80 MWe / four-pack 320 MWe) 🔥 hot
Opportunity: Industrial process-heat plus hyperscaler power are simultaneously chasing the same Xe-100 platform — Amazon for data centers, Dow for steam at Seadrift — validating SMR as a non-utility OEM business model.
Most commercially de-risked U.S. SMR right now: dual anchor customers across hyperscale and chemicals, Series C closed, and X-energy's hiring tilts to construction/supply chain/plant ops which is consistent with a project entering build phase rather than design.
VOYGR SMR module — small modular reactor ↗ rising
Opportunity: Only NRC-certified U.S. SMR design, but flagship CFPP project cancelled when subscription targets failed at ~$89/MWh — VOYGR needs a creditworthy hyperscaler/industrial offtaker to replace lost municipal demand.
VOYGR is the most regulator-mature SMR in the U.S., and NuScale is still hiring program engineers and contract managers, but it lacks an anchor customer after CFPP — risk is demand-side, not technology-side.
Natrium reactor — sodium fast reactor + molten-salt storage ↗ rising
Opportunity: Natrium's molten-salt storage gives it a dispatchable peaking advantage worth ~155 MWe for grids facing data-center spikes, but commercial start slipped 2 years because the entire U.S. HALEU supply chain was Russian-dependent.
Differentiated load-following SMR with the Wyoming demo as anchor; hiring is supply-chain and QA heavy, consistent with a project crossing from design to procurement. The binding constraint is upstream HALEU, not reactor demand.
HALEU fuel — high-assay low-enriched uranium feedstock for advanced reactors ↗ rising
Opportunity: Every Gen-IV SMR (Natrium, Xe-100, Aurora) needs HALEU and only one U.S. cascade is producing it at kg-scale — the chokepoint that already slipped Natrium 2 years is now the single biggest schedule risk to the entire SMR build-out.
HALEU is the upstream pinch point that determines whether the SMR cohort can hit late-2020s COD dates; ADVANCE Act passage reflects Congress recognizing this. Centrus is the only named domestic supplier in evidence — capacity expansion or a second supplier is a clear opportunity.
Susquehanna behind-the-meter nuclear PPA (Cumulus / AWS) — co-located nuclear-to-datacenter sleeved PPA → steady
Opportunity: The FERC 2-1 rejection and subsequent ISO tariff rulemaking define whether behind-the-meter nuclear-to-datacenter is even legal — the resolution of this proceeding is the single largest swing factor for SMR siting strategy and Talen's data-center-development hires suggest they're betting it gets resolved favorably.
Highest-stakes regulatory test case in U.S. nuclear-power right now; Talen's parallel 'Sr. Director Digital Infrastructure Development' and 'Power Generation Development' postings show the company is doubling down on co-location even with FERC headwind. Outcome here directly templates SMR-plus-datacenter projects.
Three Mile Island Unit 1 / Crane Clean Energy Center restart — large-LWR restart with dedicated hyperscaler PPA → steady
Opportunity: Microsoft's 20-yr offtake of an entire restarted reactor (fact 6c20ef89, 98c86692) creates a template for hyperscaler-anchored nuclear that Constellation can replicate across its 22 GW fleet — and is the model SMRs will need to follow.
Not an SMR but the deal that revalues the entire SMR thesis: proves hyperscalers will sign 20-yr PPAs at premium pricing for 24/7 carbon-free MW. Constellation hiring is sales-led, consistent with monetizing the rest of the fleet under similar contracts.
See the Products and Hiring modules for the full landscape and who's investing in which direction.