The massive global investment in Artificial Intelligence (AI) infrastructure is encountering significant logistical hurdles, primarily related to power supply. Despite unprecedented capital commitments from major technology corporations, nearly half of the planned data center constructions across the United States this year are expected to be postponed or shelved. Analysts attribute this slowdown to critical shortages in necessary electrical components and complex parts sourced from China.
Investment Commitments Versus Reality
Major players, including Alphabet, Amazon, Meta, and Microsoft, are projected to invest more than $650 billion by 2026 solely to expand AI processing capacity. However, according to reports from Bloomberg, these substantial financial outlays do not guarantee immediate construction progress. The core issue stems from the dependency on robust power delivery equipment—including transformers, switchgear, and batteries—which are essential both at the data center sites themselves and within the surrounding grid infrastructure required by the AI firms.
Furthermore, the electrical grid is facing added strain from rising adoption of electric vehicles (EVs) and electrified residential heating systems. Market intelligence firm Sightline Climate estimates that 12 gigawatts (GW) of data center capacity are scheduled to become operational in the U.S. in 2026; yet, constraints mean that only about one-third of this total power capacity is currently undergoing active construction.
The Critical Role of Electrical Infrastructure
While electrical components constitute less than 10% of the overall capital expenditure (CapEx) for a data center, their importance cannot be overstated. A delay in any single element along the power chain can halt an entire construction project. The expanded demand has dramatically increased lead times for high-power transformers across the U.S.: delivery periods that typically lasted between 24 and 30 months before 2020 now stretch to potentially five years, according to data cited by Sightline Climate.
This extended waiting period poses a particular threat to AI deployment cycles, which operate under an aggressive timeline of less than 18 months. Due to this intense demand, companies are increasingly turning toward international markets for vital supplies, positioning Canada, Mexico, and South Korea as key suppliers of high-power transformers for the AI sector.
Geopolitical Sourcing and Future Constraints
The supply chain remains complex and subject to geopolitical volatility. Although trade tensions have limited server manufacturers operating within China, Beijing maintains its status as the world’s largest producer of electrical equipment needed for building power infrastructure both inside and outside AI data centers. Data from Wood Mackenzie indicates a significant surge in high-power transformer imports from China, increasing from fewer than 1,500 units through 2022 to over 8,000 units by October 2025.
In terms of component sourcing, the People’s Republic of China accounts for more than 40% of U.S. battery imports, and its share in certain categories of transformers and switchgear remains close to 30%. Analysts caution that even massive investments—even those measured in trillions of dollars—will fail to translate into actual AI capacity unless the underlying power infrastructure constraints related to transformers, switchgear, and batteries are successfully resolved. The insufficient domestic manufacturing capacity for electrical gear, coupled with ongoing U.S.-China tensions, means that advanced AI data center deployments remain heavily reliant on international imports.
