Layer 1 · Where the power comes from
Nuclear PPA (existing plants)
Operators large enough to sign a multi-year offtake. Amazon (Susquehanna, 1.92 GW), Microsoft (Three Mile Island restart) and Meta (Vistra) have all done exactly this.
The shape of it
- Lead time
- 1-3 years to contract; the plant already exists
- Capital cost
- None — it is a contract, not a build
- Carbon
- Effectively zero operational carbon
- Maturity
- proven
Verified 2026-09-06.
The catch
There is a fixed supply of existing reactors and the good ones are being contracted now. This is a land-grab with a finite end, not a repeatable strategy.
What it actually decides
The fastest route to large, firm, carbon-free power that exists today — precisely because you are buying something already built.
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
- Nuclear PPA (existing plants) vs On-site natural gas turbines — Clean and slow against fast and dirty. The defining 2026 trade, and most operators are quietly choosing gas.
- Enhanced geothermal vs Nuclear PPA (existing plants) — The two clean firm options. One is geology-limited, the other supply-limited.
Others in where the power comes from
- Grid interconnection — 4-5 years typical, approaching 5 on average
- Small modular reactors (SMR) — 2030s for meaningful capacity
- 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