Power hardware shortages are delaying AI data centre expansion, despite record investment

US AI data-centre expansion is increasingly being constrained not by chips, servers or funding, but by the electrical hardware needed to connect new facilities to reliable power, Bloomberg reports. While the US–China trade war has pushed many server makers to move production out of China, the deeper dependency remains in power-delivery equipment.

China is still the world’s largest producer of electrical gear used to build and upgrade power infrastructure, both inside data centres and across the wider grid. Shortages of key components, especially transformers, switchgear and batteries, sourced from China and elsewhere, are now slowing project timelines.

The scale of planned build-outs is colliding with these supply limits. Bloomberg cites forecasts that Alphabet, Amazon, Meta and Microsoft will spend more than $650bn in 2026 to expand AI capacity, yet close to half of the planned US data-centre builds this year are expected to be delayed or cancelled.

The problem extends beyond the data-centre fence line. Companies must also fund and coordinate grid upgrades to supply enough electricity, competing for the same scarce equipment as utilities coping with rising demand from electric vehicles and electrified heating.

Sightline Climate data cited by Bloomberg suggests about 12GW of US data-centre capacity is expected to come online in 2026, but only around a third of that capacity is currently under active construction due to multiple constraints. Electrical infrastructure may represent less than 10% of total data-centre cost, but it is schedule-critical, because delays in any link of the power chain can halt an entire project.

Lead times for high-power transformers, in particular, have deteriorated sharply, typically 24 to 30 months before 2020, but now stretching to as long as five years, clashing with AI deployment cycles that can be under 18 months.

To cope, developers are turning to global suppliers, with Canada, Mexico and South Korea becoming major sources of high-power transformers. Even so, US imports of Chinese high-power transformers have surged from fewer than 1,500 units in 2022 to more than 8,000 units through October 2025, according to Wood Mackenzie data cited by Bloomberg. China also supplies over 40% of US battery imports and remains near 30% in some transformer and switchgear categories, underscoring continued reliance despite tariffs and security concerns.

Why does it matter?

Bloomberg’s central warning is that without easing bottlenecks in transformers, switchgear and batteries, and expanding US manufacturing capacity, trillions of dollars of AI investment may not translate into delivered AI capacity, because power infrastructure, not compute, is becoming the limiting factor.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!