Meta Swings to Utility-Style Heavy Debt and $145B Capex Target in Aggressive AI Pivot

Meta Platforms is undergoing a structural transition from an asset-light digital advertising firm into a capital-intensive infrastructure builder, absorbing unprecedented costs as it rushes to construct the physical backbone required for artificial intelligence.

The social media giant reported a 28% increase in second-quarter revenue, outperforming market expectations. However, a 55% surge in total costs pushed operating income down 8% and net income down 14%. The divergence between top-line expansion and profitability highlights the immediate financial toll of the company’s massive computing hardware investments.

Free cash flow for the quarter plunged to $784 million—a stark departure from the nearly $12 billion quarterly average maintained over the preceding two years. Operating cash flow reached $31.9 billion, but capital expenditure consumed $31.1 billion, representing nearly double the amount spent during the same period last year. Virtually all generated operational cash was redirected into acquiring servers, specialized processors, real estate, and data center facilities.

To sustain its buildout, Meta raised its full-year capital expenditure forecast to between $130 billion and $145 billion. Having already spent $50.9 billion during the first six months, the revised guidance implies quarterly capex will average between $39 billion and $47 billion in the second half of the year. Because projected quarterly capex exceeds typical operating cash flows, the company faces potential negative cash flow periods ahead.

To fund this long-term physical footprint, Meta raised $24.9 billion through long-term debt issuance while completely pausing share repurchases, which had exceeded $10 billion a year prior. Chief Financial Officer Susan Li stated that the company is deliberately shifting toward a higher debt balance to align financing with long-life infrastructure projects.

This heavy borrowing and asset accumulation mirror traditional utility models, where heavy upfront capital expenses, high depreciation, and debt servicing precede revenue generation. Second-quarter depreciation and amortization expenses rose 46% year-over-year to $6.4 billion as hardware assets rapidly age.

The strategy places Meta on a distinct trajectory compared to hyperscale technology peers such as Microsoft, Amazon, and Alphabet. Those companies cushion their infrastructure spending by renting out server capacity through established commercial cloud platforms. Driven by growth in data center energy demand and server rental revenue—a trend highlighted by the International Energy Agency in global energy and digital infrastructure assessments—cloud providers can generate immediate cash flows from their capital outlay. Microsoft, for instance, saw its stock rise in extended trading following strong cloud segment results.

Meta, which operates primarily consumer-facing apps including Facebook, Instagram, WhatsApp, and Messenger, lacks a mature public cloud division. Within its core Family of Apps segment, operating income contracted to $23.4 billion from $25.0 billion in the prior-year period despite higher ad revenues.

During a call with Wall Street analysts, Meta Chief Executive Mark Zuckerberg confirmed that the company is considering entering the cloud infrastructure rental market, promising details in an upcoming update. He noted that Meta receives significant third-party offers to purchase compute capacity at a premium relative to its acquisition costs.

However, Zuckerberg categorized compute leasing as a secondary opportunity, arguing that deploying computing power into internal proprietary AI services yields vastly higher profit margins over time. He told investors that selling raw compute capacity for short-term profit would be shortsighted compared to selling higher-level AI intelligence.

Meta’s target product suite includes AI systems that improve advertising conversion efficiency—which the company reported has increased by 15.7%—automated business messaging agents, model application programming interfaces (APIs), and personalized conversational assistants capable of managing user personal data. To support these development pipelines, Meta expects computing constraints to persist, with Chief Financial Officer Susan Li noting that demand for compute capacity continues to exceed available infrastructure.

Addressing long-term investor concerns regarding capital efficiency and returns through 2027, Zuckerberg maintained that high-volume spending remains necessary to secure market leadership, stating his conviction that early aggressive infrastructure investors will ultimately achieve superior returns.