AI Data Center Demand Creates 12 GW Global Capacity Deficit

A severe structural shortage in global data center capacity has emerged as artificial intelligence adoption accelerates. Jefferies Group reports that in 2025, only 8.9 gigawatts of capacity became operational against a demand of 21.1 gigawatts, leaving a 12 gigawatt deficit. This gap has pushed cloud service backlogs to approximately 2 trillion dollars.

Major hyperscalers, including Amazon, Microsoft, Google, and Meta, are driving this surge with combined capital expenditures estimated to reach 770 billion dollars in 2026. McKinsey & Company projects that total global spending on data centers will hit 4 trillion dollars by 2030. North America is expected to lead power requirements, accounting for nearly 19 gigawatts of the projected 30 gigawatt global demand in 2026.

The shortage is attributed to supply chain bottlenecks rather than a lack of investment. Critical constraints include engineering labor, cooling equipment, electrical components, and power infrastructure. AI facilities require significantly more physical inputs than traditional centers, such as up to 50,000 tons of copper per facility. This demand has increased the stock value of primary input providers, including GE Vernova for power grid components, Micron Technology for memory, and Caterpillar for construction equipment.