The world is rightly entering a 20‑year period of major expansion in nuclear power

A rapid reversal of fortune

Nuclear power has flipped from pariah to priority. Once shunned after Three Mile Island, Chernobyl and Fukushima, governments have pivoted sharply: at COP28 a “Triple Nuclear” declaration set a target to triple global nuclear capacity versus 2020, and by subsequent climate talks more than 30 countries had signed on. China has quietly become the largest construction site — 31 reactors under construction — and its domestically developed Hualong One (华龙一号) design is now in scaled deployment. International agencies project installed capacity could top 1,400 GW by 2050, up from roughly 390 GW today. The question is no longer whether nuclear matters, but how fast the world can rebuild an industry long left to rust.

Drivers: grid security, climate targets and AI

The logic is threefold. Nuclear is a controllable, low‑carbon baseload — a complement to weather‑dependent wind and solar whose intermittency has exposed grids and driven up storage costs. Nuclear plants run at very high capacity factors and can sit near major loads, lowering grid investment needs. A new and powerful driver is AI. Data centres need ultra‑reliable, steady, low‑carbon power; small modular reactors (SMRs) offer the flexibility to serve campuses and clusters in ways traditional gigawatt plants cannot. It has been reported that major tech firms are moving into nuclear deals and SMR funding — Amazon, Microsoft, Google and Meta have all been linked to investments, long‑term purchases or SMR partnerships — underscoring that compute‑hungry AI could be a structural growth engine for nuclear demand.

Uranium, supply risk and geopolitics

Rapid growth exposes new chokepoints. Uranium demand is forecast to rise sharply — to perhaps 95,000 tonnes by 2035 and over 143,000 tonnes by 2050 under current nuclear expansion plans — while known, low‑cost resources and exploration remain limited. Supply is geographically concentrated (Australia, Kazakhstan, Russia among the largest holders), and enrichment and fuel‑cycle services are geopolitically sensitive; trade policy and sanctions could therefore reshape supply chains and alliances as much as reactor technology does. Workforce gaps and decade‑long lead times for large reactors add another constraint, making SMRs and industrial policy priorities for countries that want to move fast. Can governments, miners and utilities coordinate at the pace climate goals and AI demand now require? The next two decades will decide whether nuclear’s renaissance stabilises grids — or founders on supply and geopolitical friction.