Alphabet Q3: free cash flow -$9.07B, Gemini at 1 billion users

Alphabet's AI buildout finally showed up on the cash flow statement, and the market did not like the line item.

Alphabet reported negative free cash flow of $9.07 billion for the third quarter, after spending $51.36 billion on property and equipment against $42.29 billion of cash from operations, according to the company's Q3 2026 earnings release. Revenue rose 25% to $127.58 billion but missed the $128.22 billion analyst consensus, and adjusted earnings of $2.67 a share missed $3.02. It was Alphabet's first top- and bottom-line miss in nearly four years, and the stock fell more than 10% after hours, CNBC reported.

The AI numbers inside the same release are, read alone, the best Google has ever printed. The Gemini app has reached one billion users, which CEO Sundar Pichai called the fastest-growing product in the company's history. First-party model APIs are processing nearly 32 billion tokens a minute, more than four times the year-ago rate. Google Cloud revenue grew 103% to $30.71 billion, double the prior year, and Cloud operating income more than tripled to $11.12 billion. Search revenue rose 12% to $63.35 billion, with Pichai claiming the highest year-over-year query growth rate in five years.

So the demand story is real. The problem is what it costs to serve, and how the company is now paying for it.

The quarter where the capex bill arrived

Capital spending hit $51.36 billion in the three months to September 30 — up from $23.95 billion a year earlier, and short of the $55.09 billion analysts had penciled in, per CNBC. Trailing-twelve-month capex now stands at $159.8 billion against $19.75 billion of trailing free cash flow, a ratio that would have been unthinkable for this company three years ago.

Free cash flow has now deteriorated for three straight quarters: positive $10.12 billion in the first quarter, negative $5.86 billion in the second, negative $9.07 billion in the third. Management held full-year capex guidance at the $195 billion to $205 billion range it set in the second quarter, and finance chief Anat Ashkenazi told analysts the company still expects capex "to increase significantly in 2027." CNBC puts the trajectory at more than $200 billion this year and above $300 billion next year.

Alphabet is not paying for that from operations. Long-term debt climbed to $125.1 billion from $21.6 billion a year earlier — an increase of about 479%, per CNBC — after the company issued $28.7 billion of senior unsecured notes in the quarter, raised $4.3 billion selling Class A shares and $8.1 billion selling Class C shares through its at-the-market program, and repurchased no stock at all. A company that was buying back roughly $11.5 billion a quarter a year ago spent zero on buybacks and borrowed instead. That is the clearest signal in the whole report: the AI buildout has moved from "funded by the ad business" to "funded by the balance sheet."

What the miss actually says

The miss is small in percentage terms and large in meaning. Revenue came in roughly 0.5% below consensus, adjusted EPS about 12% below. Operating income of $40.58 billion beat last year by 30% and still came in under the $43.19 billion FactSet estimate; the operating margin expanded 1.3 points to 31.8% and was read as a shortfall. Jefferies analyst Brent Thill put the shape of it plainly on CNBC's Closing Bell: "The cloud business was the only business that beat at 103% growth. The Street wanted 100%, so it didn't beat by much, and it missed some of the high-end targets. Everything else missed."

Then there is the earnings-quality wrinkle. Other income swung to a loss of $8.22 billion, driven by an $8.66 billion mark-to-market loss on equity securities — after a $99.03 billion gain in the second quarter from stakes including SpaceX and Anthropic, per CNBC. Roughly a third of last quarter's headline profit was other companies' valuations moving. This quarter they moved back.

There is a legitimate bull case, and it is not a small one. Cloud's backlog exited the quarter at $561 billion, up from $514 billion, with about $308 billion of that expected to convert within 24 months, according to CNBC's analysis of the call. Cloud's operating margin is now above 36%. Cloud chief executive Thomas Kurian, in his first public earnings call remarks, said the unit is "seeing extraordinary demand and strong growth from both AI labs, capital markets, financial service institutions, and enterprises."

And one line in the release deserves more attention than it got: TPU system sales were "a meaningful contributor" to Cloud revenue growth. Alphabet has spent a decade treating its tensor chips as an internal cost advantage. It is now booking revenue on them, which is the same shift we described when OpenAI started buying Google TPUs as Oracle began reselling them. Silicon is becoming a product line, not just a moat.

The take

For two years, hyperscaler capex was treated as free because the ad and cloud businesses could absorb it. This quarter is where that assumption met arithmetic. Alphabet can still grow cloud at triple digits and add a billion consumer users to Gemini in the same quarter and get punished for it, because investors are no longer buying growth — they are buying a payback schedule, and Alphabet has not published one past "significantly higher in 2027."

The counter-argument that matters is not that demand is weak; the numbers say it is not. It is that Alphabet is now financing a capital program larger than most national infrastructure budgets with debt and equity issuance while its free cash flow is negative, and every quarter of that adds to the amount of future cloud revenue already spoken for. Pichai told analysts that the Carbon model shipped this week outperforms Argon "on almost every domain," which is worth reading alongside Gemini 4 Carbon's top score on the AI index: the model is winning. The bill is winning too.

One more contrast from the same evening: Microsoft's quarter was received as strong by CNBC's Investing Club, while Alphabet's was not. One company is being read as converting AI spend into returns today; the other as still building the factory. The market is now pricing that difference every three months.

What to watch: the fourth-quarter call, where Ashkenazi has promised detail on the 2027 capex number — that figure, not the Gemini user count, is what will decide whether this was a bad quarter or a repricing.

Is negative free cash flow the price of winning the AI buildout, or the first sign the model doesn't pay? Write to us via our contact page.

AMD's Helios racks are live at Meta, and AMD is buying $1B of Meta

Meta's fundraising has become a customer-alignment exercise: the two vendors selling it compute are buying shares in the buyer. AMD's first Helios rack-scale systems are now installed and running inside Meta's data centers, and AMD is putting $1 billion of its own money into the equity offering that helps pay for them.

Salience Labs raises $125M for all-optical AI data center switches

The AI buildout's bottleneck moved from the processor to the wire, and today's money followed it: a British photonics startup pulled in a nine-figure round with both Intel and AMD on the cap table, while two smaller bets went to the less glamorous end of the agent economy. Salience Labs raised an oversubscribed $125 million Series B led by Bullhound Capital, with Intel Capital and AMD Ventures both participating — two chipmakers that compete for the same accelerator sockets, funding the same…

DayOne raises $3.67B to expand its Malaysia data centers

The AI build-out's financing engine kept turning overnight: another multibillion-dollar debt package for hyperscale capacity in Johor, landing days before the developer's own Nasdaq debut. Singapore data center developer DayOne has secured about $3.67 billion in financing from seven banks to expand its hyperscale capacity in Johor, Malaysia — including a $2.4 billion offshore term loan facility alongside a ringgit-denominated Islamic facility.

Meta turned down Amodei's personal plea for compute

The chip hunt is back in the headlines — and today's edition runs from a boardroom ask at the very top of the AI industry to the economics of putting robot drivers in truck cabs. Dario Amodei personally approached Meta earlier this year to source more compute for Anthropic, and Meta declined — according to the Wall Street Journal.

Ghost sells out its first run of a $3,499 personal AI computer

A screenless box that does nothing but run your AI agents sold out its first batch hours after preorders opened — while two much bigger money moves landed overnight, one of them entirely on the companies' own word. Ghost, a 19-year-old founder's startup, sold out batch one of Core, a $3,499 screenless computer built for one job: running your personal AI agents.

SoftBank's SB Energy delays a $50B IPO as investors balk

The first public price check on the AI data-center build-out came back too high for buyers, and a Russian lab put a from-scratch model on the open market the same day. SoftBank's SB Energy has delayed the data-center IPO it had planned for this month after investors pushed back on a valuation above $50 billion, the New York Times reported Monday.

85% of Nscale's $103B in AI contracts sits with two customers

Microsoft and Anthropic account for 85% of the $103.4 billion in AI compute contracts Nscale disclosed in its IPO filing — and only $2.6 billion of that was active as of late August. The London-based data center developer filed its S-1 on September 18 for a New York Stock Exchange listing under the ticker NSCL, targeting a raise of about $3 billion.

Mazama raises $135M to drill superhot rock for AI data centers

Two stories worth your time as the day turns over: the money chasing AI's power supply is now drilling into rock hot enough to melt lead, and Tesla's in-car assistant keeps deepening its ByteDance dependency. Mazama Energy has raised $135 million to prove that the hottest rock on Earth can be a power plant for AI data centers.