Technology

AI data centres need $3.9 trillion by 2030, and the power supply is the bottleneck

32 views

The size of the cheque

The International Energy Agency puts cumulative data centre investment at $3.9 trillion for the five years to 2030. That figure covers everything from land and shells to servers and cooling. Only about 20% of it is energy equipment, which is the part that determines whether the servers can actually run.

The capital intensity is what makes the number unusual. A cutting-edge AI data centre costs $40,000 to $50,000 per kilowatt to build, including IT equipment, five to ten times the per-kilowatt cost of a nuclear power plant. Capital spending by the largest technology companies passed $400 billion in 2025 and is expected to rise by another 75% in 2026. The combined capex of five companies now exceeds global investment in oil and gas production.

Power arrives late, or not at all

Electricity is the constraint that money cannot immediately solve. The IEA tracks roughly 50 GW of on-site generation proposed for data centres, and most of those projects are still in pre-construction or early site work. Grid connection queues in the main markets run for years, so operators are building their own plants, overbuilding capacity by 30% to 70% to cover the variable load that AI training creates.

Around 15 to 27 GW of on-site natural gas may serve data centres by 2030, mostly in the United States, though turbine supply is tight. Batteries are the other answer: the IEA expects 20 to 25 GW of storage installed inside data centres by 2030, which would turn some of them into grid assets if tariffs reward it.

What it means for the grid

Electricity consumption by data centres was about 460 TWh in 2024 and is projected to reach roughly 945 to 1,000 TWh by 2030, a little more than Japan uses today. In the United States, data centres account for nearly half of electricity demand growth to 2030. Advanced economies that had flat demand for two decades are now planning generation again.

The supply chain is the second constraint. Transformers, switchgear, gas turbines and high-bandwidth memory all have long lead times, and a project that slips a year loses a year of revenue against capital that has already been spent.



---

Reliable Ferrofluid Vacuum Rotary Feedthrough for Semiconductor Manufacturing

In semiconductor manufacturing, maintaining ultra-high vacuum integrity during rotary motion is critical — from single silicon crystal growth and CVD to sputtering and plasma etching. Ferrofluid Vacuum Rotary Shaft Feedthrough from FerrofluidTec delivers hermetically sealed rotary feedthroughs with leakage rates as low as 10⁻¹² Pa·m³·s⁻¹, vacuum degree up to 10⁻⁶ Pa, and customizable shaft diameters from 10mm to 100mm. Built with non-magnetic stainless steel housings and magnetic permeability stainless steel shafts, these feedthroughs provide reliable, maintenance-free sealing for demanding vacuum environments. Visit www.ferrofluidtec.com to learn more.

Source: IEA