Alphabet is running a roughly $200 billion capital expenditure program this year and President Ruth Porat told Fortune’s Most Powerful Women Summit on Monday that next year’s outlay will “increase significantly,” framing the spend as a response to persistent capacity constraints rather than a discretionary bet. The company lifted its 2026 capex ceiling to as much as $205 billion in July, a disclosure that sent the stock down more than 6% the following session, a market signal that investors are pricing execution risk, not just optionality.
The spending trajectory
Second-quarter capex hit $44.9 billion, with the vast majority directed at technical infrastructure for AI, according to CFO Anat Ashkenazi. Porat, who moved from the CFO seat to president and chief investment officer in 2023 after a decade running Alphabet’s finances and a prior stint as Morgan Stanley’s CFO, oversees the capital arms GV and CapitalG as well as the Other Bets portfolio. Her argument is structural: the 2016 pivot from “mobile first” to “AI first” created the demand pipeline that now justifies the buildout, and the spend itself is the payoff of that earlier strategy.
The cash flow inflection
The scale of outlay has rewritten Alphabet’s cash generation. Free cash flow turned negative $5.9 billion in the June quarter, the first sub-zero print in more than twenty years as a public company. At the same time, Google Cloud revenue rose 82% to $24.8 billion and operating profit more than tripled to $8.8 billion, suggesting the infrastructure is already converting to high-margin revenue. The question is whether the conversion rate sustains once the current capacity constraint eases.
The bear case and the rebuttal
Jefferies’ Chris Wood has argued the AI boom ends in massive U.S. capital destruction as cheaper open-source Chinese models erode pricing power. Porat dismissed the logic: a capacity-constrained market with strong enterprise and government demand, she said, cannot simultaneously offer no path to return. Her framework for an attractive return rests on two levers, the value added to partners to drive revenue growth, and the efficiency of compute delivery. Both are internal metrics; the market has no independent verification.
The long game
Porat’s definition of winning avoids market-share rankings. “Anything that is not built for the long run won’t endure for the long run,” she said, adding that data centers act as job creators and spread fixed energy costs across a larger base. The rhetoric is consistent: invest before the need is visible, because “by the time you need it, it’s too late.” Whether the $200 billion-plus annual run rate represents a moat or a sunk-cost trap will not be settled by next quarter’s cloud numbers.