What happened
US artificial intelligence (AI) infrastructure development faces significant bottlenecks despite substantial investment, with projected spending reaching $800 billion in 2026 and an estimated $1.16 trillion by 2027. Goldman Sachs forecasts US data centre power demand will more than double to 66 GW by 2027, up from 31 GW in 2025. This rapid expansion encounters supply chain lead times of up to four years for critical electrical components and $64 billion in delayed US initiatives due to grassroots resistance and environmental zoning issues. In contrast, China's National Development and Reform Commission is preparing a $295 billion plan for its own nationwide AI buildout. AI capital expenditures contributed 1.1% to US GDP growth.
Why it matters
The escalating demand for AI infrastructure risks significant cost increases and project delays for procurement teams and platform engineers. The mechanism involves critical component shortages, such as switchgear and transformers, facing lead times up to four years, directly impacting deployment timelines and budgets. Goldman Sachs' projection of US data centre power demand reaching 66 GW by 2027 highlights a severe constraint on grid capacity, potentially limiting new project activations. This follows previous reports that AI compute spending outpaces GDP metrics, indicating a growing disparity between investment and foundational resource availability.
