Data center electricity demand roughly doubles to ~945 TWh by 2030 on IEA numbers; Goldman pegs the growth at +165% versus 2023. The hyperscalers have already decided nuclear is the answer for firm, clean power — that part of the debate is over.^{[13] [15]}

The problem is the clock. Gigawatt-scale nuclear takes 10+ years and ~$20B per plant. Most SMRs in development rely on exotic fuels, unproven coolants, or supply chains that don't exist yet — credible delivery dates start in the 2030s. And the fast fossil alternative is gone: gas turbines are effectively sold out through 2030.^{[2] [33]}

Customers with 2028–2035 power needs — data centers, industrial heat users, utilities — have demand today and no nuclear option that ships on their timeline.^[2]

May 23, 2025. Four executive orders set a national target of quadrupling nuclear capacity from ~100 GW to 400 GW by 2050. EO 14300 ordered the NRC to decide new-reactor applications within 18 months and license renewals within 12 — with a wholesale rewrite of its regulations on an 18-month deadline.^{[8] [9]}

March–April 2026. Part 53 — the first technology-inclusive, risk-informed licensing framework — became a final rule, effective April 29. A companion proposed rule lets the NRC directly leverage DOE-authorized test reactor results in commercial reviews. The agency is already running real dockets on 9–17-month schedules, and TerraPower received the first commercial advanced-reactor construction permit in decades.^{[11] [12] [25] [44]}

One year in. DOE's own retrospective: 11 reactor pilot projects selected, the DOME test bed open at Idaho National Lab, $2.7B committed to restoring enrichment capability, and 31+ countries signed onto tripling global nuclear capacity by 2050.^{[10] [39]}

A commercial PWR steam generator stands up to 70 feet tall, weighs 300–800 tonnes, and contains thousands of hand-assembled U-tubes; a plant needs two to six of them. They are among the largest and most expensive hand-built components in all of industry — and their size dictates the containment building, the construction schedule, and ultimately the decade-long timeline.^[35]

Apollo's Compact Steam Generator replaces that vessel with a component 20x smaller at the same thermal power — per the launch video, "thousands of needles through 5% of the space" — cutting the plant's overall footprint by ~80% in volume terms and making the full system roughly 40x more compact: pre-tested, factory-built, transportable, and contracted at a fixed price.^{[2] [4]}

Everything else stays inside the NRC's comfort zone: light water, standard low-enriched uranium, fuel and components from existing qualified supply chains, and the PWR architecture backed by 15,000+ reactor-years of global operating experience.^{[2] [7]}

Apollo filed its Regulatory Engagement Plan with the NRC on April 2, 2026 (ADAMS ML26092A282, pre-application docket 99902160) — publicly searchable, signed by Halimi, and specifically scoped to licensing the A-10. The plan targets a construction permit application by 2028.^{[5] [6]}

Two topical reports on the steam generator technology are planned for submission this year, built on A-0 and A-1 test data. Because the CSG is the only novel component, the licensing surface area is a fraction of what an entirely new reactor type carries — the NRC has approved PWR physics dozens of times.^[2]

The sober counterweight: prediction markets price a new-reactor license being granted in 2026 at just 24%. The EO-driven clocks are young, and the 2028 target leans on them holding. Apollo's hedge is design conservatism — when reviews drag, exotic designs wait longest.^[40]

Fuel. Most funded SMR challengers — Oklo, X-energy, TerraPower, Radiant, Antares — need HALEU, a fuel the US has produced roughly 920 kg of, total, against multi-tonne per-reactor needs. Domestic enrichment covers an estimated 10–25% of projected 2050 demand, and the shortage already pushed TerraPower's Natrium out two years. Apollo's standard low-enriched uranium ships commercially today from Framatome and Westinghouse.^{[26] [31] [32] [43]}

Components. BWXT — sole manufacturer of nuclear components for the US Navy, with a $6B backlog — and Framatome are under commercial agreements, covering pressure vessels, fuel, and instrumentation. Siemens Energy (turbines) and ABB (controls) are under memoranda of understanding. Nothing in the bill of materials requires a factory that doesn't exist.^[34]

Turbines. Steam turbines carry a 2–3 year backlog — tight, but a different world from gas turbines, which GE Vernova expects to be sold out through 2030 with ~3-year lead times. Apollo's stated goal is compressing typical 5–10 year nuclear vendor lead times to 2 years for first delivery.^[33]