Meta Q2 earnings miss as AI costs surge; capex raised to $130–145B

Meta's Q2 2026 earnings reveal a widening tension at the heart of its investment thesis. Revenue continues to climb—up roughly 28% year-over-year to about US$60.80 billion—demonstrating that the core advertising business remains resilient. Yet profitability has contracted sharply: net income fell to around US$15.85 billion, and the company has now raised its 2026 capex guidance to US$130–145 billion, signaling that AI infrastructure spending is accelerating rather than moderating. This puts investors in a difficult position: they must believe that Meta's extraordinarily expensive data center and AI model buildout will eventually unlock new revenue streams and justify the near-term margin compression, or they must accept that the company is overinvesting in infrastructure whose returns remain uncertain.

The El Paso data center partnership with BlackRock—a US$14 billion, 1 gigawatt commitment—exemplifies this tension. By structuring the investment off-balance-sheet, Meta gains scale without immediately burdening its cash flow statement, but the underlying capital commitment is real and substantial. The company is doubling down on AI infrastructure at precisely the moment when investors are most skeptical: the earnings miss (driven by cost growth outpacing revenue growth) and the capex guidance hike have already sparked a 6.6% stock decline. Analysts are diverging sharply on what 2029 will look like. The baseline narrative assumes Meta achieves $369.0 billion revenue and $111.2 billion earnings by 2029—a 19.7% compound annual revenue growth rate—but more pessimistic observers project only US$343.5 billion revenue and roughly US$99.8 billion earnings, suggesting the AI payoff will be weaker. Which forecast prevails may depend less on Meta's execution and more on whether the broader market believes that large-scale AI infrastructure investments can generate durable competitive advantages and profitable new use cases.