Despite tech giants planning to spend over $650 billion on AI infrastructure in 2026, close to half of all planned U.S. data center builds this year are projected to be delayed or canceled. The culprit isn't a lack of money or computing chips. It's something far more mundane but absolutely critical: electrical equipment .

The bottleneck centers on transformers, switchgear, and batteries that power both data centers and the electrical grids feeding them. These components represent less than 10% of a data center's total construction cost, yet a delay in any single element can halt an entire multibillion-dollar project. It's a classic case of infrastructure constraints outpacing investment capacity .

Why Are Electrical Components in Such Short Supply?

The problem stems from a perfect storm of global factors. China remains the world's largest producer of electrical equipment needed for power infrastructure, yet U.S.-China trade tensions have made relying on Chinese suppliers increasingly risky. At the same time, demand for electrical components has exploded not just from AI data centers, but also from the broader electrification of transportation and heating systems .

The numbers tell a stark story. Before 2020, high-power transformers typically arrived within 24 to 30 months. Today, waiting periods stretch to as long as five years, according to market intelligence firm Sightline Climate cited in recent reporting . For AI companies operating on deployment cycles of under 18 months, this is catastrophic.

To illustrate the scale, approximately 12 gigawatts of data center capacity is expected to come online in the U.S. in 2026. Yet only about one-third of that capacity is currently under active construction due to various constraints, with electrical infrastructure being the primary culprit .

How Are Companies Adapting to the Supply Crisis?

Facing these constraints, tech companies and their partners are pursuing several strategies to keep projects moving:

• Diversifying suppliers: Canada, Mexico, and South Korea have become major suppliers of high-power transformers for AI data centers, reducing reliance on any single source.
• Increasing Chinese imports despite tensions: Imports of high-power transformers from China surged from fewer than 1,500 units in 2022 to more than 8,000 units through October 2025, suggesting companies are willing to navigate geopolitical risks to secure equipment.
• Accepting higher costs: The volatility and scarcity of components are driving up prices, with China accounting for over 40% of U.S. battery imports and maintaining roughly 30% market share in certain transformer and switchgear categories.

Despite a decade of reshoring initiatives aimed at reducing U.S. dependence on foreign manufacturing, domestic electrical equipment production remains insufficient to meet current demand .

What Does This Mean for the Future of AI Infrastructure?

The electrical equipment shortage represents a fundamental constraint on AI expansion that money alone cannot solve. Even with trillions of dollars in capital and cutting-edge computing hardware ready to deploy, actual AI capacity will depend on power infrastructure availability, not capital or compute constraints .

Looking ahead to 2026 and beyond, the industry is undergoing a broader transformation in how it approaches data center design. AI-ready data centers are evolving into highly specialized facilities engineered to support extreme compute, power, and thermal demands. This includes a shift from traditional air cooling to liquid cooling systems that can handle rack densities exceeding 50 to 100 kilowatts, compared to the 5 to 10 kilowatts typical of legacy enterprise racks .

Power engineering is becoming the critical path for infrastructure projects. According to projections cited in industry analysis, data center electricity demand could rise by over 165% by 2030, with AI workloads driving much of this surge . This means that grid-to-chip power engineering, from high-voltage distribution to on-site substations and battery energy storage systems, is no longer a secondary consideration but a primary design driver .

The convergence of supply constraints, geopolitical tensions, and unprecedented demand growth means that infrastructure limitations are beginning to define the pace of AI innovation itself. Without resolving bottlenecks in transformers, switchgear, and batteries, the AI revolution's physical foundation remains fragile, regardless of how much capital is deployed.