Demand is arriving before supply

Artificial intelligence data centres need enormous amounts of electricity around the clock. Wind and solar are cheap but intermittent, and gas is difficult to reconcile with emissions targets, so nuclear offers stable, emissions-free power 24 hours a day. That AI-driven demand is the starting point for renewed interest in nuclear and small modular reactors. It is also creating a fuel problem that financing alone cannot solve.

Tech deals are power deals

In autumn 2024 Microsoft signed a 20-year agreement with Constellation that could pave the way for restarting Three Mile Island Unit 1. Google concluded an agreement with Kairos Power for 500 MW of next-generation reactors. Amazon invested in X-energy with a target of 5 GW of capacity within 15 years. In January 2026 Meta announced agreements with Vistra, TerraPower and Oklo that could finance up to 6.6 GW of capacity by 2035, including up to eight sodium-cooled reactors.

Publicly available information does not indicate that any of the tech giants directly procured nuclear fuel. A plant operator signs separate contracts for uranium, conversion, enrichment and fuel assembly fabrication, usually over several years. Tech companies buy electricity, so the entire fuel process falls to operators and reactor developers. For young companies such as TerraPower or Oklo, that means building a supply chain from scratch.

TerraPower's fuel warning

In December 2022 TerraPower announced that commissioning of its first sodium-cooled Natrium reactor in Wyoming would be delayed by at least two years. The delay was not caused by lack of financing, technology problems or regulatory obstacles. It was caused by lack of fuel. The initial plan was to buy the first batch of HALEU from Russia, and the invasion of Ukraine made that impossible. At the time, outside Russia, no one produced HALEU at industrial scale.

An energy supply contract with a tech giant can be signed in a few months, but building an enrichment plant takes years. The lag from deposit discovery to mine start-up is currently 10 to 20 years.

After the 2011 Fukushima disaster, uranium prices stayed low for more than a decade and investment in new mines and enrichment plants ceased to be profitable. New demand is hitting a market that had been shrinking for years. The further up the supply chain, the harder it is to ramp production quickly, and the less media attention the constraint receives. That gap between market attention and physical supply may be the most important investment asymmetry in the nuclear fuel cycle.

This article was assisted by AI analysis. Please refer to the original source for official information.