Massive AI datacenter CapEx

- McKinsey projects $5.2 trillion in cumulative AI and cloud CapEx by 2030 under its base case, with a $7.9 trillion high‑case that anticipates even larger buildouts. - The report flags power as the bottleneck: U.S. transmission capacity is growing ~2% while data‑center and AI load could increase 15–20%, forcing whole‑site upgrades. - That mismatch pushes spending beyond GPUs into power, cooling and grid upgrades — a structural demand tailwind for infrastructure vendors. (x.com 1) (x.com 2)