AI & ComputeLOT 261119 · 2 min read
Schneider Warns Power Prices Could Drive AI Investment Out of Europe
Schneider Electric's digital chief warns high power prices could push AI investment out of Europe, which holds under 5% of global compute and may need up to 145 TWh for AI by 2030.
Key takeaways
- Europe holds less than 5% of the world's computing infrastructure, per Schneider's November 2025 study.
- Schneider modeled European AI power demand at 45–145 TWh by 2030 across three scenarios.
- Schneider is lead energy infrastructure partner on a SoftBank-led plan for up to 5 GW of AI data centers in France.
- Weckesser spoke to Reuters in Gdansk on Nov. 19.
- US leads data center construction, China follows, Europe lags; Middle East expanding faster.

Europe holds less than 5% of the world's computing infrastructure, and its high electricity prices could push AI investment to other regions. That is the warning from Peter Weckesser, chief digital officer of Schneider Electric, speaking to Reuters in Gdansk on Nov. 19.
Weckesser's argument is simple. Energy costs feed directly into the cost of computing. High power prices, insufficient grid investment and fragmented regulation across the European Union could slow the region's AI data center expansion, he said.
The competitive picture he describes is not flattering. The United States leads data center construction, followed by China, while Europe lags, according to Reuters. Regions including the Middle East are expanding faster than Europe. Weckesser said Europe needs major grid investment and a more consistent regulatory framework.
How much power does European AI need?
Schneider, which sells power distribution, cooling and control equipment to data center builders, put its own numbers on the problem in a November 2025 study. The figures cover European AI electricity demand by 2030 under three scenarios:
- 45 TWh in a constrained scenario
- 90 TWh in a coordinated scenario
- 145 TWh under unchecked expansion
The study called for grid capacity to be built ahead of demand, not behind it. A grid that arrives late is a cost, not an asset.
Schneider is not a bystander in this market. The company is the lead energy infrastructure partner on a SoftBank-led plan for up to 5 GW of AI data center capacity in France. It sells the equipment that the AI buildout requires, which makes its demand modeling more than academic.
Should Europe control the AI supply chain?
On calls for European control of the AI supply chain, Weckesser was blunt. Full control is unrealistic and would raise costs.
"I think it's much smarter to think about which parts of the supply chain you really need to own," he said, "and which ones you are able to source from the outside."
The comment cuts against the current of European industrial policy, which has leaned toward sovereignty in semiconductors, energy and cloud. Weckesser, whose company profits from global supply chains, suggests selectivity over autarky. Ownership of everything is ownership of every cost overrun.
What is at stake for data center builders?
The commercial logic is straightforward. If electricity in Europe costs more than elsewhere, and the grid cannot deliver capacity when needed, hyperscalers will place their next gigawatt elsewhere. The Middle East is already expanding faster, per Reuters.
The 2030 horizon gives Europe roughly five years to close a gap that spans generation, transmission and regulation. Schneider's own study frames the downside at 45 TWh — the constrained scenario — versus 145 TWh if expansion runs unchecked. That spread is, in effect, the price of policy failure.
For now, the US remains clearly ahead in construction. Whether Europe's answer arrives as grid investment, regulatory consolidation or simply slower AI growth will determine where the next wave of data center capital lands.
mlq.ai (Source)
More from Nathan Brooks
Show full bio
Senior reporter covering people, launches and day-to-day practice in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.
1,132 articles