Amazon buys stake in nuclear energy developer in push to power data centres

Original Article ↗ Hacker News Discussion ↗

Scope of the announcement

  • AWS talked about plans for ~5 GW of small modular reactors (SMRs), joining other big tech nuclear moves (Google, Microsoft, Oracle, Sam Altman, Gates, etc.).
  • Some see this as a meaningful financial commitment with real downside risk; others note current investments are small compared to the eventual multi‑tens‑of‑billions needed.

Nuclear vs. solar/wind + storage

  • Pro‑nuclear side:
    • Argues AI/data centers need gigawatt‑scale, 24/7, low‑carbon power that wind/solar can’t reliably provide due to intermittency.
    • Claims batteries can’t economically cover multi‑day or “arbitrary length” weather shortfalls; storage requirements would explode as renewable penetration rises.
    • Notes existing nuclear fleets run at ~93–96% capacity factor and can be planned for refueling/maintenance, unlike weather.
  • Pro‑renewables side:
    • Cites rapidly falling costs and huge global build‑out of solar, wind, and batteries; nuclear additions are tiny by comparison.
    • Points to modeling showing fully renewable systems (with flexibility and storage) cheaper overall than nuclear‑heavy systems, with nuclear needing ~80–85% cost reduction to compete.
    • Argues demand can be shifted and grids can handle intermittency with forecasting, interconnection, demand response, and diverse sites.

Reliability and grid behavior

  • One camp emphasizes nuclear’s very high fleet‑wide capacity factors and uncorrelated outages, calling it the most reliable large‑scale generation.
  • Critics counter with:
    • Long unplanned outages at individual plants.
    • Examples where many reactors are simultaneously offline.
    • Issues like heat‑wave cooling constraints and political shutdowns.
  • Consensus: nuclear is not 100% available; renewables are highly variable. Disagreement is about which mix gives the most dependable system at lowest cost.

Economics and subsidies

  • Nuclear advocates: once built, plants “print money”; fuel is cheap; long lifetimes amortize capex. They blame nuclear troubles on policy, stop‑start build programs, and distorted markets (priority, subsidies, and negative prices for renewables).
  • Skeptics: highlight cost overruns (Vogtle, Hinkley, Flamanville), bailouts and renationalizations, special tariffs, and underfunded waste management as evidence nuclear isn’t truly profitable without heavy state support.
  • Both sides accuse the other of ignoring system‑level costs (transmission, backup, curtailment, waste, decommissioning).

Climate and technology strategy

  • One faction: “If you’re not serious about nuclear, you’re not serious about climate,” citing Germany’s nuclear exit vs. France’s low‑carbon mix.
  • Opponents say Germany is decarbonizing via renewables and that market trends overwhelmingly favor solar/wind+batteries; nuclear’s slow deployment (10–15+ years) makes it marginal for urgent climate timelines.
  • Alternatives discussed: hydro, pumped storage, geothermal, tidal, advanced reactors (breeders, molten salt, thorium), but their scalability and maturity are debated.

Data centers and power

  • Strong synergy argued between nuclear and data centers: steady, high, year‑round load; shared expertise in high‑reliability complex systems.
  • Others note DCs still need grid interconnection and that, today, GPUs are far more expensive than electricity, so even costly grid power plus renewables can make economic sense.
  • Some foresee a split: nuclear‑backed “elite” power for AI and industry vs. cheaper, variable renewables for general consumers.