Layer 1 · Where the power comes from
Small modular reactors (SMR)
Operators planning baseload past 2030. Google (Kairos Power), Amazon (X-energy) and Meta (Oklo, TerraPower) have signed.
The shape of it
- Lead time
- 2030s for meaningful capacity
- Capital cost
- Very high, and first-of-a-kind cost risk is real
- Carbon
- Effectively zero operational carbon
- Maturity
- emerging
Verified 2026-09-06.
The catch
Nuclear belongs in the 2030+ baseload layer, not the 2026 energisation path, unless you inherit an existing plant interconnection. Any vendor implying otherwise is selling a press release.
What it actually decides
A real answer to the decade-out question and no answer at all to the this-year question. Both things are true and the industry keeps conflating them.
What this is powering
A GB200 NVL72 rack draws 120–132 kW and a GB300 pushes 135–200 kW, against a 2026 average rack of about 27 kW. The silicon decision and the power decision are the same decision, made eighteen months apart.
Layer 2 — Silicon · GB200 NVL72 · The 150 W inference option
Weighed against
- Small modular reactors (SMR) vs Grid interconnection — Two answers to the same question, both arriving in the 2030s.
Others in where the power comes from
- Grid interconnection — 4-5 years typical, approaching 5 on average
- Nuclear PPA (existing plants) — 1-3 years to contract; the plant already exists
- On-site natural gas turbines — 12-24 months, the fastest firm power available
- Behind-the-meter generation — Depends on the generation, but it skips the queue
- Solar + battery storage — 18-36 months
- Wind PPA — 24-48 months