Meta’s AI spending surges to $31B, cratering Q2 free cash flow

Meta's pivot from a capital-light software business to a capital-intensive hardware and AI infrastructure power is straining its balance sheet. The collapse in free cash flow signals that the company is effectively reinvesting almost every dollar of ad profit back into GPUs and data centers to maintain its competitive position. For the streaming and video ecosystem, this massive technical moat limits the ability of smaller players to compete in AI-driven content recommendation and synthesis. Investors must now weigh Meta’s 28% top-line growth against the reality of zero stock buybacks and mounting debt as it finances a speculative superintelligence roadmap. Watch for the August trial outcomes, where state attorneys general seek penalties that could further drain reserves.

The broader technology sector is facing an 'AI Capex Dilemma' as infrastructure spending outpaces internal cash generation. Per the Financial Times and Morgan Stanley (July 2026), Microsoft, Amazon, Alphabet, and Meta are projected to deploy approximately $725 billion in capital expenditures this year, a 77% increase from 2025. This spend is increasingly reliant on external financing; incremental annual debt for these hyperscalers rose to 32% of capex in mid-2026, up from just 9% in 2024. FactSet reporting in July 2026 highlights that free cash flow is moving toward zero for almost all major players except Alphabet and Microsoft due to these front-loaded costs.

Meta’s financial strain is compounded by escalating legal threats regarding platform safety. Per JURIST (July 2026), a coalition of state attorneys general is seeking up to $1.4 trillion in penalties in an upcoming August trial, alleging the company intentionally designed addictive features for underage users. This figure closely mirrors Meta's total market capitalization, representing an existential regulatory risk. Additionally, a New Mexico jury recently hit Meta with a $375 million verdict in July 2026 for failing to stop child exploitation, providing further momentum for over 1,000 pending school district lawsuits across the U.S.

Strategically, Meta is transitioning its AI efforts from the open-weight Llama series toward proprietary frontier models. In April 2026, Meta Superintelligence Labs released Muse Spark, a closed-weight reasoning model designed to power its next generation of AI-enabled smart glasses and business agents. While CEO Mark Zuckerberg maintains that AI is already driving advertising returns via products like Advantage+, the shelved release of the 2-trillion-parameter 'Behemoth' model suggests technical hurdles remain as the company scales its mixture-of-experts architecture.