Meta's AI Gamble: Rent Out Power or Build Own Empire?

Meta is navigating a critical juncture, balancing its ambitious AI development with the potential to monetize its vast computing infrastructure. The company's second-quarter free cash flow plummeted by 91% year-over-year, reaching just $784 million, a stark indicator of its significant investment in AI.

CEO Mark Zuckerberg views computing power as a scarce, strategic asset crucial for building Meta's future AI models and services. Despite acknowledging lucrative offers to rent out this capacity, he emphasizes retaining it for internal use. This strategy contrasts with competitors like Microsoft, which have successfully leveraged AI investments to boost cloud and assistant services.

Investors remain cautious, drawing parallels between the current AI expenditure and the company's previous costly and largely unremunerative pivot to the metaverse. Meta's capital expenditure forecast has been raised, reflecting an ongoing commitment to AI build-out, even as financial scrutiny intensifies.

While Meta's CFO, Susan Li, asserts that industry-wide underinvestment in AI capacity creates significant return opportunities, the ultimate profitability of Meta's compute strategy, whether through internal use or external rental, remains a key question for the market.