The Nuclear Energy Comeback: Climate Math or Wishful Thinking?

Constellation Energy announced in September 2024 that it would restart the Three Mile Island Unit 1 reactor (the unit that did not melt down in 1979), renamed Crane Clean Energy Center, under a 20-year power purchase agreement with Microsoft. The deal — the first long-term nuclear PPA of this type — was driven by Microsoft's need for large-scale clean power for its AI data centers. The same month, Amazon and Google announced nuclear energy deals.

Why Nuclear, Why Now

The revival case for nuclear is primarily climate and reliability driven. Nuclear provides firm, carbon-free baseload power — generation that is available 24/7 regardless of weather. In electricity grids with high renewable penetration, the intermittency of wind and solar creates need for either energy storage (currently expensive at scale) or firm low-carbon generation. Nuclear fills this role.

Data center power demand — driven largely by AI training and inference — has created acute demand for large quantities of clean, reliable electricity. Data center operators (Microsoft, Amazon, Google) have made climate commitments and face grid carbon accounting that makes nuclear attractive even at premium prices. The willingness of a Fortune 500 company to sign a 20-year nuclear PPA at premium rates may be the financing breakthrough the nuclear industry has needed.

The Economic Reality

The structural economic challenge for nuclear is high upfront capital costs and long construction timelines. Overnight construction costs for new nuclear in the US run approximately $6,000-$9,000 per kilowatt of capacity, compared to roughly $1,000-$2,000/kW for utility-scale solar and wind. The construction timeline for large light-water reactors in Western countries has averaged 10-15 years, during which capital earns no return. Georgia's Vogtle Units 3 and 4 — the first new US reactors in decades — were completed in 2024 at approximately 2.5x budget and several years late.

Small Modular Reactors

The industry's near-term hope is Small Modular Reactors (SMRs) — reactor designs in the 50-300 MW range, intended to be factory-built (reducing on-site construction complexity), deployed at lower absolute capital cost per unit, and scalable. The theory is that factory production creates learning curves and cost reductions that large custom-built reactors cannot achieve.

NuScale Power received NRC design certification for its 77 MW SMR design in 2023, the first SMR design certified in the US. However, NuScale cancelled its first planned project (Carbon Free Power Project in Idaho) in November 2023 due to cost overruns in the project's economics. Other SMR developers — TerraPower (Gates-backed), X-energy, Kairos Power, Oklo — are in earlier development stages.

Fusion's Role

Fusion energy — which has been "30 years away" for 70 years — has attracted significant private investment in recent years. Commonwealth Fusion Systems (MIT spinout), TAE Technologies, Helion Energy (which has a conditional PPA with Microsoft), and over a dozen others have collectively raised several billion dollars. The December 2022 National Ignition Facility result (net fusion energy gain) provided scientific validation. Commercial fusion power by 2035 is now a seriously held position in parts of the investment community — though still highly uncertain.