Two-thirds of a trillion

Add up what Microsoft, Alphabet, Amazon, Meta, and Oracle plan to spend building AI infrastructure in 2026 and the figure lands near $690B — up from about $443B a year earlier. The headline shorthand making the rounds, that the group is 'nearly doubling' its spend, is only true firm by firm: Amazon and Alphabet each roughly doubled their own budgets, but the five-company total rose about 55%. The real story isn't the growth rate. It's that a single year's construction budget for five companies now sits at two-thirds of a trillion dollars, against a Gartner forecast of $2.5T in total worldwide AI spending for 2026.

Capital spending used to take about 40 cents of every operating dollar these firms earned. In 2026 it takes nearly all of them.

What changed is where the money comes from. For most of the past decade these firms spent roughly 40 cents of every operating dollar on capital projects and banked the rest. At 2026's pace, capital spending consumes nearly all of their operating cash flow, and the group issued more than $121B in bonds in 2025 to cover the gap — holding, for the first time, more debt than cash. The build-out has outgrown the profits funding it.

The strangest note comes from the forecaster itself. Gartner places AI in its 'Trough of Disillusionment' for all of 2026 — the phase where early hype curdles and the technology gets quietly sold by existing vendors rather than chased as a moonshot — even as it projects spending to hit $2.5T. An industry can be underwhelming its believers and setting capital-spending records in the same year. Whether that ends in durable infrastructure or a write-down is the question every dollar is betting on.