Google's Q3 2026 Earnings Miss: Why AI Costs Are Eating Profits

Alphabet reported third quarter results on Wednesday, and the stock dropped about ten percent after hours. Revenue grew twenty-five percent, and investors still hated it.

Twenty-five percent growth and the market punishes you? That's a company the size of a small country growing like a startup.

It is. A hundred and twenty-seven point six billion dollars in revenue. But it missed estimates slightly, and profit missed badly. Net income fell to twenty-six point two billion, down from nearly thirty-five billion a year ago.

So profit dropped a quarter while sales grew a quarter. Something's eating the middle.

Two things. The AI buildout, and one ugly one-off. Adjusted earnings were two dollars sixty-seven a share against three oh two expected. Reported earnings were two twelve, dragged down by an eight point two billion dollar unrealized loss on Alphabet's stake in SpaceX.

Hang on, SpaceX? That's a mark-to-market thing, not the business, right?

Right, paper loss, nothing to do with search or cloud. Strip it out and you still have the real problem. Free cash flow was NEGATIVE nine point one billion dollars.

Google burned cash? Google is the company that prints it.

For twenty-two years as a public company it never had a negative quarter. Last quarter was the first, five point nine billion burned. This quarter it's nine. Analysts expected roughly break-even.

Okay, so where's the money going? Data centers, I assume.

Capital spending of fifty-one point four billion dollars in the quarter alone. The full-year forecast is a hundred and ninety-five to two hundred and five billion, which they raised by fifteen billion back in July. And the CFO, Anat Ashkenazi, told analysts capex will increase significantly again in twenty twenty-seven.

Significantly meaning what? Did she give a number?

No number. She said details on the Q4 call. CNBC reports the expectation is annual capex topping two hundred billion this year and exceeding three hundred billion next year.

Three hundred billion. On what, concrete and chips?

Chips, buildings, energy. And they're financing it. Long-term debt went from twenty-one point six billion a year ago to a hundred and twenty-five billion now.

That's a sixfold jump. Google borrowing money is genuinely strange to me.

One investor quoted by Reuters, Jeremy Saltsburg at Check Capital, put it this way: a three hundred billion capex number could scare the market, because free cash flow is finally negative.

Fine, but spending is a choice. You spend because you expect returns. Is the money coming back?

That's the real question of the quarter, and the answer is split. Google Cloud more than doubled, to thirty point seven billion dollars. Their backlog of signed commitments is five hundred and sixty-one billion, and they expect to book more than half of that as revenue within twenty-four months.

That sounds enormous. So why is everyone upset?

Because search decelerated. Ad revenue came in at eighty-two billion, missing by about one point four billion. Search and other ads grew twelve percent, the slowest in five quarters according to Bloomberg.

And search is what pays for all of it.

It's roughly half the company. Chief business officer Philipp Schindler blamed AI Mode, their chatbot-style search interface. More people are using it, and it monetizes worse than classic search results.

So Google's own AI is eating Google's ad business. That's an uncomfortable sentence to say on an earnings call.

He said they feel very good about the trajectory, that ad experiments show promising results, and they're scaling ads inside AI Mode to more users and more markets.

Does anyone outside the company buy that?

Mixed. Gil Luria at D.A. Davidson told CNBC the burden of proof is now on Google to show it isn't in permanent decline as people shift to chatbots. Dave Wagner at Aptus said investors need to see search accelerating, not decelerating, and cloud is carrying the story.

What would the bull case even be here? Spend now, monetize later?

Essentially. Cloud demand is real, they're now selling their own TPU chips as full systems, and Thomas Kurian said that's become a meaningful contributor. But building TPU inventory pressures cash flow further, Ashkenazi warned, along with margin pressure from renting third-party cloud capacity.

Wait, Google rents compute from someone else? They have the data centers.

They can't build fast enough. And the forward commitments are staggering. By the end of the quarter, future obligations to buy compute and energy were nine hundred and twenty-four billion dollars, plus a hundred and fifty-two billion in data center leases that haven't started. Over a trillion, total.

A trillion dollars of bills arriving later, against an ad business that just slowed down. I see why the stock moved.

Thomas Monteiro at Investing.com said until Google shows it can charge for AI the way it charged for search ads, every capex forecast reads less like ambition and more like a bill.

So what settles it? The next quarter?

The Q4 call is the one to watch. That's when Ashkenazi promised the twenty twenty-seven capex number. Watch whether search advertising reaccelerates as they put ads into AI Mode internationally, and whether cash flow turns positive.

And if it's three straight negative quarters?

Then the argument stops being about ambition. For now, revenue grew twenty-five percent and nobody cared.

Strange week to be the most profitable search engine on earth.