Meta's Free Cash Flow Falls 91% as AI Infrastructure Spending Surges

Meta Platforms has reported a sharp fall in free cash flow as the Facebook and Instagram owner accelerates spending on the data centres, servers, chips, networking systems and power infrastructure required for artificial intelligence.

The company generated just $784 million in free cash flow during the second quarter, down from approximately $8.55 billion in the same period one year earlier. That represents a year-on-year decline of about 91%.

Free cash flow measures the money remaining after a company pays for its normal operations and capital expenditure. Investors watch the figure closely because it indicates how much financial capacity is available for dividends, share buybacks, debt reduction, acquisitions and further expansion.

Meta's operating business continued to generate substantial revenue, supported by digital advertising across Facebook, Instagram and its wider group of applications.

However, the company's capital expenditure rose significantly as chief executive Mark Zuckerberg pursued an ambitious AI strategy.

Meta is building large-scale computing capacity to train and operate increasingly advanced artificial-intelligence systems. Zuckerberg has argued that computing infrastructure will remain a strategically valuable and scarce resource.

The company expects the investment to support more capable consumer assistants, advertising tools, content-recommendation systems and automated services for businesses.

Meta has also suggested that excess computing capacity could eventually be provided to outside organisations, potentially creating an additional source of revenue.

Investors remain uncertain about how quickly those opportunities will generate returns sufficient to justify the scale of spending.

The company's shares declined following the results as markets reacted to the fall in cash generation, higher projected expenses and questions about AI monetisation.

Meta reported that quarterly revenue rose strongly, but second-quarter profit declined after significant legal and restructuring expenses.

The debate surrounding Meta reflects a wider question across the technology industry: whether enormous investment in AI infrastructure will produce durable profits or create excess capacity before demand is fully developed.

Microsoft, Amazon, Google, Oracle and other technology groups are also committing large sums to AI data centres and specialised processors.

Unlike cloud-computing companies that can sell processing power directly to corporate customers, Meta has historically relied heavily on advertising. It must therefore show that AI will either increase advertising efficiency, create successful new products or produce an entirely new revenue stream.

The company argues that AI is already improving recommendations and advertising performance. Nevertheless, the collapse in free cash flow demonstrates that the infrastructure build-out carries a considerable near-term financial cost.

Meta remains highly profitable and continues to generate strong operating cash flow. The central concern is therefore not immediate solvency, but capital allocation: how much the company should spend, how quickly it should build and when shareholders can expect measurable returns from the investment.