What happened
RAND published this quantitative research report on 25 June 2026, mapping the full inventory of critical electrical equipment required for AI data centre operations and stress-testing U.S. power grid capacity against supply chain failure scenarios out to 2030. The headline finding: supply chain threats to front-of-the-meter components (natural gas generation, storage, transformers) have the potential to result in 'a decrease of approximately 7 percent to 31 percent in available net capacity by 2030' relative to a no-delay baseline, while BTM battery supply chain threats could cut another 8 percent of net capacity. Using a composite supply chain vulnerability score, the authors identified steam turbines, geothermal production wells, and conductors and wires as the most vulnerable FTM equipment in 2025; for BTM, backup power components dominate the risk ranking. A key structural finding is that 'data centers cannot eliminate supply chain constraints by moving off-grid' because off-grid bridge-power installations compete for the same scarce natural gas turbines as grid-connected projects. The report recommends that the Department of Energy adopt a tiered policy decision framework, that the U.S. government prioritise generation-system supply chain interventions over transmission, and that trade authorities increase Harmonized Tariff Schedule granularity for the most vulnerable components.
Why it matters
AI infrastructure investment plans premised on continuous power availability face a materially underappreciated risk: the U.S. electrical equipment supply chain is structurally constrained in ways that could cut AI-grade grid capacity by up to a third by 2030 — a finding that should directly inform board-level capital allocation decisions on data centre expansion, real-estate site selection, and energy procurement strategy.
Action needed
Commission a supply chain risk review of your AI data centre energy roadmap against the RAND vulnerability taxonomy, with particular focus on transformer and natural gas turbine lead times; brief the CFO and COO on the 7–31% capacity shortfall scenario before committing to new AI infrastructure capex.