Meta’s AI bet rattles investors as spending soars without clear payoff

Meta

Meta is spending heavily to stay ahead in the AI race — but investors are beginning to wonder what exactly the company is buying.

As part of its massive AI buildout, Meta is constructing two large data centres and pouring billions into infrastructure and talent. Reports suggest the company could spend as much as $600 billion on US infrastructure over the next three years. While such numbers might not shock Silicon Valley, they have unsettled Wall Street.

The concern came to light during Meta’s latest earnings report, which showed operating expenses jumping by $7 billion year-on-year and nearly $20 billion in capital costs. These increases stem largely from AI investments that have yet to generate tangible revenue. When pressed for clarity, CEO Mark Zuckerberg insisted that the spending is necessary to ensure Meta has the computing power to train frontier models and push the company into what he called “a massive latent opportunity.”

That reassurance did little to calm investors. Meta’s stock dropped 12% by Friday’s close, wiping out more than $200 billion in market value. Despite posting $20 billion in quarterly profit, the company’s first visible hit from AI-related spending has raised doubts about its direction. Beyond vast data centres and a swelling payroll of AI researchers, there’s little evidence of a product delivering financial returns.

Zuckerberg faced pointed questions from analysts about how and when these investments might pay off, but offered only broad statements about future AI-driven experiences and improvements across Meta’s family of apps. His remarks suggested that the company is still in a transitional phase, with no clear revenue stream tied to its AI initiatives.

That uncertainty sets Meta apart from rivals. Google and Nvidia are spending billions too, yet both continue to deliver strong quarters backed by clear AI monetisation strategies. Even OpenAI — despite its heavy losses — can point to ChatGPT’s explosive growth and a multibillion-dollar annual revenue run rate. Meta, by contrast, has no breakout AI product anchoring its ambitions.

The Meta AI assistant, which Zuckerberg says now reaches over a billion users, benefits from integration into Facebook and Instagram but still falls short of being a true ChatGPT competitor. Other experiments, such as the Vibes video generator, have boosted engagement but offer limited business impact. The recently launched Vanguard smart glasses also appear more like a continuation of Meta’s Reality Labs work than a commercially viable AI product.

Zuckerberg remains optimistic, pointing to the company’s new Superintelligence Lab and its next generation of models. “We expect to build novel models and novel products,” he told analysts, promising more details “in the coming months.” But as the market reaction made clear, investors are losing patience with promises of future breakthroughs.

Only four months have passed since Meta reorganised its AI division, so it may be too early to judge results. Yet as billions continue to flow into AI infrastructure, the question remains: what is Meta’s real play in artificial intelligence? Whether it’s a consumer chatbot, an enterprise platform, or something entirely different, the company needs to prove that all this spending is building more than just potential.