Microsoft plans to more than triple data center capacity to 38 GW
Microsoft's data center footprint is about to get unrecognizable. According to a Bloomberg report citing people familiar with the plans, the company intends to expand its global data center capacity from roughly 12 gigawatts today to more than 38 GW by 2032 — an addition of about 26 GW, or more than triple the current base. About a third of that 38 GW would be dedicated to racks of AI-specific chips, a signal of just how much of Microsoft's future build-out is aimed squarely at training and serving frontier models rather than ordinary cloud workloads.
For scale: 38 GW is a serious chunk of a mid-sized country's power grid, and the AI-focused third alone rivals the 14.8 GW Anthropic signed for just last week. Microsoft is making the single largest infrastructure bet any company has made on the assumption that AI demand keeps compounding through the decade. If demand softens even a little, this plan becomes the most expensive excess capacity in corporate history; if it holds, whoever has the megawatts owns the market. Bloomberg's reporting also suggests the build-out isn't optional — The Information's write-up on the same plans carries a telling phrase in its headline: Microsoft is "hurt by server shortage," meaning it is already capacity-constrained while demand outruns supply.
We covered Anthropic's 14.8 GW compute signing last week — Anthropic has signed for 14.8 GW of compute — and maybe $517 billion — and Microsoft's plan shows the hyperscalers are responding in kind.
Nvidia's CEO says 70% growth next year is in the bag. Speaking at Goldman Sachs' Communicopia + Technology conference, Jensen Huang argued Nvidia "could grow 70% year over year" — analysts already expect roughly $400 billion in revenue this fiscal year, so that forecast implies somewhere near $680 billion next. His confidence rests on visibility: "We're tracking every single gigawatt of land, power, shell around the world," Huang said, because every lab, cloud, and neocloud reports back to Nvidia. On circular-financing concerns, his answer was characteristically blunt: "We put in $1 and $100 comes back in. Is that circular? If that is, let's do more of that." Huang's chart and Microsoft's 38 GW plan are the same bet made twice — the entire AI economy now rises or falls on whether this compute actually gets used.
What to watch: whether Microsoft confirms the 38 GW figure on the record — and which utilities sign the power contracts that make it real.
If a third of Microsoft's 38 GW is AI-only, what happens to cloud prices for everything that isn't AI?
