科技巨头AI基建狂潮无人叫停

After four of the biggest competitors in the AI race held quarterly earnings calls late last month, two things are clear: Big Tech is not done spending record amounts on the AI data center buildout, and the AI data center buildout is not any closer to yielding its promised returns.

A third thing is also starting to become clear — Big Tech will not be slowing down, at least not on its own.

Amazon, Microsoft, Alphabet, and Meta told investors they’re planning to spend at least $720 billion on CapEx this year alone. Most of that will go toward AI infrastructure. That’s on top of at least $1 trillion in spending commitments that have been arranged for the coming years.

Big Tech’s spending plans officially spooked Wall Street and triggered a sharp selloff in tech stocks last week that began to taper off as investors shifted focus to Tuesday’s midterm elections. Those spending plans could change after voters across the country elected candidates who favor moratoriums and other restrictions on data centers.

Wall Street might see the new uncertainty around the future of data center regulation as a good thing, given that tech stocks began to rise again on Tuesday. After all, there’s another issue Big Tech companies have to address. Despite enormous amounts of spending, they haven’t really made money from AI yet, and they’re continuing to double down on it anyway.

Until now, investors haven’t really blinked as Big Tech has ramped up its spending to fund the AI buildout. But more than three years in, Wall Street is starting to get antsy about the promised payoff of AI — and exactly when they might expect to see a return on investment.

It’s not that top-line revenue isn’t growing — it very much is. Amazon reported its fastest growth in five years for the quarter, Meta’s ad revenue is surging, and Google Cloud’s revenue more than doubled in a year. But as astonishing as it sounds, it might not be enough to justify the massive amounts Big Tech is spending on AI.

Alphabet offered the most uncomfortable example of this last week when Philipp Schindler, Google parent Alphabet’s senior vice president and chief business officer, acknowledged on the company’s earnings call that Google Cloud’s revenue growth has come at the expense of its core Search business.

“It’s worth noting that the strong growth in AI Mode queries had an impact because of the lower monetization we were seeing in AI Mode,” Schindler said.

Revenue from Google’s Search ads grew 12% in the third quarter, a marked drop from the 17% growth rate the division posted for the same quarter last year.

It’s a weird paradox for Google. The strength of the company’s core business, Google Search, is what previously gave investors the confidence in the company’s aggressive spending agenda for AI. But now, Google’s new AI-enabled search engine is eating into theSearch business before the company has really figured out how to monetize AI Mode.

“To be clear, we feel very good about our trajectory here. Our ad experiments have shown promising results,” Schindler said of Google’s efforts to monetize AI Mode. He did not offer any specific examples.

Wall Street was especially unhappy with Google last week: On Thursday, Alphabet stock saw its worst one-day drop in history, sinking 13% in a single day.

The stock has started to pick back up this week, but it probably isn’t helping that the only thing Big Tech can say with certainty so far is to expect to see more spending.

Alphabet, which plans to spend between $195 billion and $205 billion in capital expenditures this year, is also planning to increase spending “significantly” in 2027, according to CFO Anat Ashkenazi.

Amazon intends to spend “meaningfully more” in 2027 than the $220 billion in capital expenditures it is planning this year, CEO Andy Jassy said on the company’s earnings call last week. He framed the spending — the largest ever in the company’s history — as a “once in a lifetime opportunity” to grow Amazon into a $1 trillion company.

“Would it be more convenient if this type of acceleration and opportunity happened over 25 years and you could spread the CapEx over 25 years? Yeah, it might be more convenient,” Jassy said. “But that’s not the world we live in, and that’s not the opportunity.”

Meta, which is expected to spend between $135 billion and $145 billion this year, didn’t disclose how much it intends to spend next year, but major financial institutions are expecting it to increase its spending from this year.

In touting the company’s rollout of Muse, Meta CEO Mark Zuckerberg noted that the agentic AI program will only become profitable if Meta can get its compute costs down.

Zuckerberg noted that Meta expects compute efficiency to increase “tenfold” next year, an endeavor he expects will take “significant investment.”

It’s not just Big Tech that is burning cash. Anthropic updated its S-1 prospectus with third-quarter revenue this week, revealing a sharp uptick in computing costs and net losses. The AI startup’s revenue grew to $17 billion in the quarter, a more than 1,000% increase from last year, but its compute costs came in at $17.5 billion — not quite enough to cover the impressive gains.

Anthropic has also committed nearly $900 billion in future compute and infrastructure costs, and it plans to begin the costly endeavor of building its own data centers.

Some of that $900 billion will go to Google, which is selling its TPU chips to Anthropic. Two different analysts on Google’s call pressed executives on the TPU business. After all, Google, like the other hyperscalers, has said it is capacity-constrained. If that is true, why would Google spend its resources selling its own capacity to other labs?

When asked if the money made from selling TPUs would help Google tamper some of its own spending, Ashkenazi didn’t give a clear yes or no answer. However, the answer seems likely to be no. Google is fronting the costs of building the hardware it sells to AI labs like Anthropic. Those contracts are supported by financial obligations that will come due when the hardware is ready to be used, but they don’t represent cash in Google’s pocket right now.

It may be worth noting that Microsoft was the only Big Tech company to escape Wall Street’s ire last week. The company posted $45.2 billion in capital expenditures for the quarter, down from an expected $50 billion. However, the lesser spend wasn’t due to any austerity on Microsoft’s part, but rather “third party capacity delays.”

Until last week, Microsoft had been the only hyperscaler that hadn’t taken on debt to fund its AI buildout. “We are fortunate to have access in the debt markets, given where our ratings are, and we have not yet, I would say, tapped that,” CFO Amy Hood said on the company’s earnings call Wednesday night.

That had changed by the next morning, when news of Microsoft’s $25 billion, 10-part bond sale broke.

By the time of this writing, Big Tech stocks were back up. Maybe Wall Street is banking on the “blue wave” of the midterms to force Big Tech to slow down some of its spending in the coming years. A slowdown might also give Big Tech a little more time to figure out how to make AI profitable.

Then again, despite the midterm elections looming last week, they did not come up on any of Big Tech’s earnings calls.