What happened
Bain's 2026 Technology Report analysis on AI data-centre supply projects that meeting AI compute demand requires structural, system-level moves rather than site-by-site workarounds. Using its Data Center Model, Bain projects '$5 trillion to $6.5 trillion of buildout, adding about 150 gigawatts or more by 2030. That would nearly triple global capacity over five years' — with power, semiconductors, skilled labour, and permitting all simultaneously constrained. It notes local opposition blocked or delayed 'at least 75 projects worth $130 billion in the first quarter of this year alone,' and argues the gap will only close via consolidation, new power at scale, orbital data centres, government investment, or new social licence — warning that 'executives planning on business as usual will get blindsided.'
Why it matters
Quantifies the AI compute supply shortage and its multi-year lead times, giving boards the $5–6.5T capex and constraint picture needed to stress-test AI infrastructure and vendor lock-in decisions.
Action needed
Test your AI procurement and capacity plans against the 2030 supply gap and long lead times for grid, chips, and permitting identified in the model.