Google just ran out of free cash for the first time in a decade, on purpose

Alphabet just posted one of its best quarters in years. It also ran its cash flow negative for the first time in a decade, on purpose, and investors punished the good news anyway.

By almost every normal measure, Alphabet's second quarter of 2026, reported on July 22, was a strong one. Revenue climbed 24% year over year to $119.8 billion, beating the roughly $116.5 billion analysts expected. Earnings per share came in at $9.11, far ahead of the $2.88 consensus. Operating margins expanded to 34%. Google Cloud, the company's enterprise AI infrastructure business, grew revenue 82% to $24.8 billion, with a backlog that reached $514 billion and existing customers exceeding their contracted spending commitments by more than 50%.

Alphabet's stock fell 7% anyway. The reason wasn't the quarter that had just happened. It was the one still to come.

The number that actually moved the market

Alongside the results, Alphabet raised its full year capital expenditure guidance to $195 to $205 billion, up from a prior range of $180 to $190 billion, almost entirely to fund AI data centers, chips and infrastructure. Quarterly capex alone reached $45 billion in Q2, up from $36 billion in Q1. To help fund it, Alphabet said it would issue up to $85 billion in new stock specifically "to scale AI infrastructure and global compute."

The consequence showed up in a number Alphabet has not reported in roughly ten years: free cash flow, the cash left over after running the business and paying for investments, turned negative, at -$5.9 billion. CFO Anat Ashkanazi attributed the swing directly to the capex increase.

Why this is a stranger story than "big tech overspends on AI"

That framing is familiar and mostly wrong here. Alphabet isn't spending blindly hoping demand shows up later. Google Cloud's backlog of $514 billion and customers already spending 50% more than they committed to are unusually concrete signals of real, current demand, not speculative future demand. CEO Sundar Pichai said the company is "winning new customers, more than doubling our acquisition velocity." This is closer to a company that is capacity constrained, unable to build data centers fast enough to keep up with paying customers, than one betting on a market that might never arrive.

The actual tension is different: a company famous for decades as a cash generating machine, one whose free cash flow funded stock buybacks and cushioned every downturn, has voluntarily traded that cushion away, mid-boom, to avoid being the hyperscaler left without enough compute to sell.

A simple way to picture it

Imagine a bakery so popular that the line is out the door every morning, and instead of quietly enjoying record sales, the owner takes out a large loan to buy a second oven, a walk in freezer and a delivery van, all at once, because turning away customers today means losing them to the bakery across the street tomorrow. The spending isn't reckless. It's a bet that under-building costs more than over-building, in a market moving this fast.

"We're winning new customers, more than doubling our acquisition velocity." - Sundar Pichai, Alphabet CEO, on Google Cloud's growth

The honest catch

The market's nervousness isn't baseless. Analyst Chris Ballard described the sector as "transitioning from AI hype to monetization discipline," meaning investors now want proof that this capex converts into durable profit, not just growth. Alphabet's own backlog and cloud growth numbers are the strongest such proof available among the major AI infrastructure spenders right now, but a $514 billion backlog is a promise of future revenue, not cash already banked, and multi-year AI infrastructure bets carry real risk if enterprise AI adoption slows or if cheaper compute alternatives emerge before Alphabet's spending is repaid. A negative free cash flow quarter is sustainable for a company with Alphabet's balance sheet. It would be a five alarm fire for almost anyone else.

Alphabet just proved its AI business works well enough to sell out its own capacity. The market's real question isn't whether the demand is genuine. It's whether spending faster than you earn, even for good reasons, is a habit any company, even this one, can sustain indefinitely.