Big Tech's AI spending outran cash by $33B for the first time
What's the deal? Five companies driving the data centre build-out — AlphabetDealroom has a profile for this one. Try Dealroom →, Amazon, Meta, Microsoft and Oracle — spent $33 billion more on capital projects last quarter than they generated from operations. It marks the first time their combined free cash flow has turned negative, a milestone in the AI infrastructure boom.
Why now? The build-out has rested on the balance sheets of some of the world's wealthiest companies. With their latest quarterly results, that model reached a tipping point: the cash burn ran heavier than anticipated, and a wave of new debt and equity issuance — including $95 billion of stock from Meta — underlined a growing reliance on outside financing.
The numbers in context: ReutersDealroom has a profile for this one. Try Dealroom → data shows the five spent $556.9 billion on capital expenditure across the first nine months of 2026 against $542.9 billion in operating cash flow — the first such shortfall since records began in 2011. This is not a sign the core businesses have stalled: operating cash flow rose 29% year on year.
What changes for investors? Less cash comes back to them. Dividends and buybacks fell to 9.3% of operating cash generated, down from 27.4% a year earlier.
What's the endgame? The companies show no signs of slowing. Several have said they plan to raise capital spending further in 2027, on top of the more than $800 billion they are expected to invest this year.
What could go wrong? The giants' finances rest on entrenched cash cows that AI could itself disrupt. A Google executive said in October that its new AI-powered search generates less advertising than traditional search — a comment that wiped a record $500 billion off Alphabet's market value in a single day.
The scale of what's coming: One forecast puts the cost of the AI build-out at some $5 trillion by the end of 2030, $4 trillion of which could be debt. That is equivalent to the entire US municipal bond market, according to Tomasz Tunguz, a venture capitalist at Theory VenturesDealroom has a profile for this one. Try Dealroom →.
The signal: The AI build-out has become a capital-markets story. Stijn Van Nieuwerburgh, a professor at Columbia Business School, estimates planned facilities will account for an average of 3.6% of GDP annually through 2032 — economically larger than the 19th-century railroad frenzy. The central question for investors is whether returns will justify spending that executives say will keep rising.
Read more: Financial Times
Image credit: U.S. Army Combat Capabilities Development Command
