Meta Q2 2026: Revenue Up 28% but Stock Plunges 9% as AI Capex Balloons to $145B

Meta Platforms reported its strongest revenue quarter in company history on Tuesday evening—and the market still punished it. The company’s Q2 2026 earnings, released after the close on July 29, showed revenue of $60.8 billion, a 28% year-over-year increase powered by advertising and AI-enhanced content ranking. But an EPS miss of nearly 14% and a sharply higher capital expenditure outlook sent shares tumbling roughly 9% in after-hours trading to around $529, after closing the regular session at $585.

The result crystallizes a growing anxiety among investors: what happens to margins when a company burns $31 billion in a single quarter building infrastructure for an AI future that has not yet shown up fully in earnings?

Where the Numbers Came In

Meta’s top line was unambiguously strong. Ad impressions grew 14% year-over-year, average price per ad climbed 12%, and total revenue landed at $60.8 billion against Wall Street’s $60.2 billion estimate—a modest beat on the top line that masked severe pressure below it.

Total expenses surged 55% to $42.0 billion, driven by three converging forces: AI infrastructure investment, legal provisions, and workforce reorganization costs. Net income fell to $15.8 billion and diluted earnings per share came in at $6.18—badly trailing the analyst consensus of $7.14 to $7.17 per share.

The miss broke down roughly like this: $2.4 billion in charges related to legal proceedings (the company did not specify which cases) and $1.2 billion in severance expenses connected to the approximately 8,000-employee layoff wave announced in May 2026. Without those items, Meta would likely have beaten EPS expectations. But the combination, added to a Q3 guidance range whose midpoint came in below consensus, was enough to trigger an aggressive sell-off.

Capital Expenditure: The Number That Spooked the Street

The single most consequential data point in the report was a revision to Meta’s full-year 2026 capital expenditure outlook. The company raised its guidance to $135 billion to $145 billion—up from the $125 billion to $145 billion range it had set just one quarter earlier. Critically, the low end of the range jumped by $10 billion in a single quarter, suggesting that AI infrastructure spending is not merely a ceiling but a floor.

In the just-completed Q2, Meta spent $31.1 billion on capital expenditures, more than double the $17.0 billion it spent in Q2 2025. Annualized, that pace would imply roughly $124 billion in full-year capex from the first half alone—consistent with the new guided range but alarming in its velocity.

Meta’s Iris custom AI chip is expected to enter production in September, which should eventually reduce the company’s dependency on Nvidia hardware and lower inference costs per unit. But in the near term, the ramp-up in infrastructure spending is outpacing the revenue contribution from AI-native products.

Zuckerberg’s Pitch—and Its Limits

CEO Mark Zuckerberg used the earnings call to argue that AI investment is already paying off in the core advertising business, pointing to improved feed ranking, more relevant Reels recommendations, and higher advertiser returns on Meta’s platforms. “AI is accelerating our core business today, powering our next generation of products, and opening the door to entirely new enterprise opportunities,” he said.

He positioned the buildout as a foundational bet: Meta is constructing the infrastructure necessary to serve one billion AI agent interactions per day across WhatsApp, Messenger, Instagram, and third-party surfaces. The company’s Muse Spark 1.1 agentic API, launched in July, is the first commercial expression of that ambition—a developer platform enabling businesses to deploy AI agents at Meta scale.

Wall Street’s skepticism is not about the vision. It is about the time horizon and the free cash flow in the interim. Meta’s free cash flow—once one of the most reliable metrics in consumer internet—has come under severe pressure as capex doubles, and Q3 revenue guidance of $61 billion to $64 billion (midpoint $62.5 billion, versus consensus $63.1 billion) suggested the top-line deceleration has not fully reversed.

The Contrast With Microsoft

Meta’s evening came immediately after Microsoft reported a quarter that was widely interpreted as the most convincing AI-investment payoff story yet among megacap tech companies. Azure’s 43% growth and 30 million Copilot seats gave Microsoft investors a clear narrative: the money spent on AI is showing up in cloud revenue acceleration and enterprise seat counts.

Meta’s challenge is different. Its AI spend is primarily serving free consumer products through advertising monetization—a model where the AI-to-revenue chain has more intermediate links and longer latency. When the story is “AI will make our ads better,” investors accept that framing at much lower capex levels. When capex hits $145 billion and the guidance is light, the same investors start asking whether the model can generate sufficient returns.

What’s Next

Meta will attempt to close the credibility gap in Q3 with its Muse and Meta Business AI platform driving early enterprise revenue. The September production start for its Iris chips is another landmark to watch—any acceleration in internal silicon deployment would meaningfully reduce the external hardware bill.

For now, though, Meta exits its Q2 with the contradictory profile that has come to define its 2026 AI arc: the strongest revenue growth among major social platforms, an AI narrative that Zuckerberg defends with conviction, and a market that wants to see the infrastructure cost curve bend before it gives the story full credit.