Meta is making one of the largest AI bets in the market, and the scale is no longer just a Silicon Valley talking point. It has become a Wall Street question: what, exactly, is the product that turns this spending into meaningful revenue?
The company’s latest earnings call put that question in sharp focus. Meta is building two massive data centers, and reporting indicates there will be as much as $600 billion in spending on U.S. infrastructure over the next three years. At the same time, investors saw AI costs begin to show up more clearly in the company’s financial results.
The bill is becoming visible
Meta reported quarterly earnings showing operating expenses jumping $7 billion year-over-year and nearly $20 billion in capital expense. The source of that spending was clear: AI talent and infrastructure.
That level of investment is meant to support research, new models and the compute needs of Meta’s existing business. Mark Zuckerberg told analysts that Meta wants to accelerate spending so it has the compute it needs for AI research, new work and improvements to its core business.
He also framed the effort as a future opportunity tied to more advanced models. The company is aiming for frontier models with capabilities that are not available elsewhere, and Zuckerberg described that as a “massive latent opportunity.”
But the call did not calm investors. Meta’s share price fell sharply after the earnings discussion, and by the closing bell on Friday, Meta’s stock dropped 12%, representing more than $200 billion in lost market cap.
Strong earnings did not answer the AI question
The market reaction was not simply about whether Meta is profitable. The company still posted $20 billion in quarterly profit, which is a major cushion. The issue is that this was the first quarter where Meta’s aggressive AI spending on talent and infrastructure visibly affected the bottom line.
That made the missing product story harder to ignore. Analysts asked why Meta was spending so much and when revenue from that spending might arrive. The answers stayed broad.
Zuckerberg pointed to new content formats, business versions of AI products, stronger recommendations across the Family of Apps and improved advertising recommendations. Those are logical places for Meta to apply AI, because they connect to the company’s existing platforms. But they do not yet amount to a clear revenue forecast anchored by a specific product.
This is the center of Meta’s AI product problem. The company can explain why AI matters, and it can point to enormous infrastructure. What it has not yet shown is a product that makes investors comfortable with the pace and size of the spending.
Why Meta is being judged differently
Meta is not alone in spending heavily on AI infrastructure. Google and Nvidia also had a great quarter, and OpenAI is spending enormous sums with less financial cushion than Meta.
The difference is that OpenAI has a clearer product story. Sam Altman can point to one of the fastest-growing consumer services in human history, bringing in $20 billion a year in revenue. That does not settle every debate about sustainability, but it gives investors a concrete business to evaluate.
Meta does not yet have an equivalent. The company’s core business may help it withstand pressure better than many others if AI spending turns into a bubble. But the question remains: what is Meta’s fast-growing AI product at the center of the investment case?
For now, the answer is unsettled. Meta has promising experiments, but they do not yet appear to carry the same weight as a major, fast-growing AI service with a clear revenue line.
Meta’s current AI products are not enough yet
The strongest existing candidate is the Meta AI assistant. Zuckerberg said on the call that it has more than a billion active users. But those users are supported by the reach of Facebook and Instagram, which have three billion active users, and the current product is hard to view as a direct rival to ChatGPT.
Meta also has Vibes, its video generator. It did boost daily active users, but the business impact described so far appears limited beyond that usage lift.
The Vanguard smart glasses are another ambitious launch. Released earlier this month, they may be an important product for Meta, but they look more connected to the company’s Reality Labs work than to a clear attempt to turn large language models into a defining business.
Taken together, these products show that Meta is experimenting across assistants, video and hardware. What they do not yet show is a fully formed AI business that explains the scale of spending.
The pressure is now on the next generation
Zuckerberg’s most telling answer was not about the products already in market. When pressed on infrastructure spending, he emphasized what is coming next from the Superintelligence Lab’s new models.
He said the company expects to build “novel models and novel products,” and that he was excited to share more when Meta has it. But an earnings call is not a product launch, and investors were left with the promise that more would be shared “in the coming months.”
That leaves several open strategic questions inside Meta’s AI push:
- Will Meta AI become a stronger competitor to ChatGPT?
- Will Vibes become part of a broader consumer entertainment strategy?
- Will “business AI” turn into a more defined enterprise play?
- Will AI mainly improve recommendations and advertising across the Family of Apps?
Any of those directions could matter. The problem is that Meta has not yet made one of them clear enough to match the scale of its spending.
It has only been four months since Zuckerberg restructured the company’s AI team, so the new Superintelligence team has not had much time to deliver a major product. Still, the market has made its concern clear. Meta is spending billions of dollars to stay competitive in AI, but investors now want to see what role the company intends to play in the industry, and how that role becomes revenue.