In the midst of an unprecedented AI ramp-up, Meta is spending more than most companies. The company is building two large data centers and reportedly expects to spend as much as $600 billion on U.S. infrastructure over the next three years.
Those numbers may not raise eyebrows in Silicon Valley, but Wall Street is starting to worry.
The issue surfaced this week when Meta reported its quarterly results, which showed that the company’s operating expenses jumped $7 billion from a year earlier, and capital expenditures reached nearly $20 billion. This was the result of significant spending on AI talent and infrastructure, which has yet to generate meaningful revenue for the company. When analysts asked for more specifics, Mark Zuckerberg said the spending was just beginning.
“The right thing to do is to accelerate this to make sure we have the compute that we need for both the AI research and the new things that we’re doing, and try to take a different state of computing stance in our core business,” Zuckerberg told analysts on a conference call. “Our view is that if we can bring the new models that we’re building at MSL into there and have something like a true frontier model with novel features that you won’t find anywhere else, this is just a huge potential opportunity.”
Even if his goal was to reassure investors, it didn’t work. By the end of the conference call, Meta’s stock price had plummeted. Even after two days, the defeat continues to deepen. Meta’s stock price fell 12% by Friday’s close, wiping out more than $200 billion in market capitalization.
It’s dangerous to read too much into stock prices, and in strict financial terms, Meta’s quarterly earnings weren’t too bad. (A quarterly profit of $20 billion is hard to argue with.) But this was the first quarter in which Meta’s aggressive AI investments in both talent and infrastructure had a measurable impact on the company’s bottom line. Even more worryingly, it was unclear what the money actually bought, aside from a number of huge data centers and highly paid AI researchers.
Analysts asked Mr. Zuckerberg why he is spending so much money on AI and when he expects to see a return on increased spending. However, this call comes at an odd place in Meta’s plans, as there is no clear budget for anticipated spending and no available products to lock in revenue projections. As a result, Zuckerberg was left with only general claims about the potential of AI.
tech crunch event
san francisco
|
October 13-15, 2026
“There’s going to be all kinds of new products based on different content formats. And we’re starting to see that,” he said aside during the call. “And then we also have business versions of all of these things, like Business A. The other part is how more intelligent models improve our core business, improve the recommendations we make across our family of apps, and improve the recommendations we make in advertising.”
Meta is not alone in spending billions on AI infrastructure. So it’s worth figuring out why the same spending from Google and Nvidia, which had great quarters, hasn’t surprised investors. OpenAI is the biggest offender, spending the same amount of money with a much smaller financial cushion than Meta.
There are real concerns that we are creating a bubble, and if one were to occur, Meta’s core business would be able to weather things better than other companies.
But if you ask Sam Altman why he spends hundreds of billions of dollars on computing, he’ll tell you that he runs one of the fastest-growing consumer services in human history, generating $20 billion in annual revenue. While we can debate how sustainable the growth rate is (we’ll discuss this in another blog post), underlying the OpenAI hype is actually a fast-growing product. The rapidly growing ARR numbers go a long way in answering the question.
Meta has no such product, and it’s not clear where it would come from.
The company’s most powerful AI product is its Meta AI assistant, which Zuckerberg said on the conference call has more than 1 billion active users. But these numbers are definitely driven by Facebook and Instagram’s 3 billion active users, making it hard to imagine that the current version of Meta AI is a competitor to ChatGPT. There’s also the Vibes video generator, which has actually increased the number of daily active users, but has limited impact on the business beyond that.
The most ambitious project is the Vanguard smart glasses, which were released earlier this month. However, the glasses feel more like an extension of Meta’s Reality Labs efforts than an actual attempt to harness the power of LLM.
Simply put, these are promising experiments, not fully formed products.
It’s telling that when pressed on infrastructure spending, Mr. Zuckerberg’s response was to focus on the next generation rather than cite recent product developments.
Although Zuckerberg emphasized that the impact of the Superintelligence Institute’s new model is pending, he stressed that he is very excited about the new product.
“It’s not just meta-AI as an assistant,” he said. “We look forward to developing new models and new products. Once we have them, we look forward to sharing more.”
But since this was an earnings call rather than a product announcement, all he could say was that he would have more to share “in the coming months.”
As the market response shows, that answer is fading.
To be fair, it’s only been four months since Mr. Zuckerberg reorganized his company’s AI team, and the new superintelligence team hasn’t had time to launch an earth-shaking AI product yet. But while the company is spending billions of dollars to stay competitive in the AI space, it’s not yet clear what role Zuckerberg wants to play in the new industry.
Will Meta AI leverage the company’s deep storage of personal data to become a competitor to ChatGPT? Will Vibes build on Meta’s targeted advertising system and become the first step in its consumer entertainment business? Or does Zuckerberg’s reference to “business AI” signal a more detailed corporate strategy?
So far, it’s anyone’s guess. Whatever the answer is, the pressure is on the meta to find it quickly.
