July 30, 2026, (Inside AI) — Meta reported a 14% drop in second-quarter profit to $18.3 billion, even as revenue surged 28% to $60.8 billion. The decline was driven by a 55% jump in costs to $42 billion, fueled by aggressive spending on artificial intelligence infrastructure.
The company raised the lower end of its annual capital expenditure forecast to $130 billion, up from $125 billion in April, with most funds directed toward AI data centers. CEO Mark Zuckerberg defended the outlays, stating AI is now accelerating every major part of the core business. He also flagged selling raw computing power as a potential new revenue stream.
Investors recoiled, sending shares down more than 9% in after-hours trading. The reaction echoes broader market anxiety over Big Tech's $1.5 trillion collective data center bill expected this year and next, following Google's own capex hike last week.
Meta's spending spree is unique among hyperscalers. Unlike Amazon, Microsoft, and Google, it lacks a cloud computing business to monetize AI infrastructure directly. In June, Anthropic offered to buy computing power from Meta in a deal potentially worth $10 billion; talks are ongoing. Zuckerberg confirmed the company is exploring selling compute as a service.
"We are now at a point where our investments in AI are accelerating every major part of our core business," Zuckerberg said. He added that "potentially selling compute directly" could be a new business opportunity.
The pivot to AI has been bumpy. This month, Meta yanked a feature letting users generate AI images of each other on Instagram days after launch due to privacy backlash. Yet it also released Muse Spark, its most advanced AI model under chief AI officer Alexandr Wang, alongside an image generator, Muse Image, with a video generator due in coming months. A more powerful model, code-named Watermelon, is slated for fall.
Muse Spark still lags rivals on coding, reasoning, and writing benchmarks, according to internal assessments. Meta's open-source approach has drawn fire from competitors, but Zuckerberg pushed back in a New York Times interview Tuesday, saying tightly controlling AI would be "abandoning our values" and stifle innovation.
Meta's legal woes persist. It spent $2.4 billion on legal fees in the quarter, partly tied to addiction litigation. It lost the first of nine bellwether trials in March but gained a reprieve when a 15-year-old Florida plaintiff dropped his suit this month.
Reality Labs, home to AI smart glasses, generated $431 million in revenue, up 16%, but lost $4.6 billion, flat year-over-year. The family of apps grew to 3.6 billion users, a 3% increase.
"People think AI spending is supposed to slow down, but who's going to be the first company to blink?" said David Wagner, head of equity at Aptus Capital Advisors.
Meta's capex trajectory underscores a high-stakes bet: build infrastructure first, find monetization later. The Anthropic talks signal a possible path, but with no cloud business, Meta must prove AI can lift ad revenue enough to justify the tab, or that selling compute becomes a viable third-party business. For now, Wall Street remains unconvinced.
