Nuclear Power Gets a Second Life as Cloud Giants Race to Feed AI’s Energy Demands
The numbers are staggering. Global electricity consumption from data centers is on track to nearly double from 485 terawatt-hours in 2025 to 950 TWh by 2030, according to the International Energy...
The numbers are staggering. Global electricity consumption from data centers is on track to nearly double from 485 terawatt-hours in 2025 to 950 TWh by 2030, according to the International Energy Agency. AI workloads are the main driver, tripling over the same period. By some projections, data centers could top 1,000 TWh by 2026—more than Japan uses in a year.
For database migration specialists, this energy crunch poses a direct challenge. Moving legacy databases to the cloud isn’t just about schema changes or latency anymore. It’s about where you put the compute. Hyperscalers like Microsoft, Amazon, and Google are no longer waiting for grid upgrades. They’re striking deals to restart shuttered nuclear plants, back small modular reactor (SMR) startups, and secure carbon-free baseload power directly.
Microsoft made the first big move, signing a 20-year agreement with Constellation Energy to restart Three Mile Island—a $16 billion bet on commercial operation by 2028. Amazon bought a data center campus next to the Susquehanna nuclear plant for $650 million, then locked in a 1.9-gigawatt power purchase agreement through 2042. Google chose the future, contracting with Kairos Power for up to seven SMRs, with first units expected by 2030.
Why does this matter for cloud migration? Because the cloud’s promise of infinite scalability hits a wall when power isn’t guaranteed. Natural gas still fills the gap today, but it clashes with net-zero pledges. Renewables add volume but can’t deliver the 24/7 high-density power AI training requires. Nuclear runs 24 hours a day with capacity factors above 90% and near-zero emissions. For enterprises moving critical databases to the cloud, the choice of provider may increasingly hinge on who can guarantee steady, affordable power.
The practical takeaway: as you plan your legacy-to-cloud migration, factor in regional energy stability. The hyperscalers are placing their bets on nuclear. If SMRs deliver on cost and schedule, they’ll reshape where data centers land—and how fast your migration can scale.