July 30, 2026, (Inside AI) — Meta's second-quarter earnings revealed a dramatic 91% collapse in free cash flow to just $784 million, driven by an AI infrastructure spending spree that CEO Mark Zuckerberg defends as essential for long-term dominance. The stock plunged 9% premarket on Thursday as investors questioned the strategy.
Zuckerberg framed compute as a scarce strategic asset, not a commodity to sell for quick profit. Yet he admitted the company had received offers for its computing capacity "at a meaningful premium" over investment costs, exposing a core tension: should Meta rent out its chips to ease cash-flow pressure or hoard them for its own AI ambitions?
The dilemma underscores Meta's awkward position. Unlike cloud giants Microsoft, Alphabet, and Amazon, which can sell excess compute directly to enterprise customers, Meta lacks a hyperscaler business model. Its AI spending must ultimately be justified by its advertising business, which still generates the vast majority of revenue.
"Meta is spending like a hyperscaler without a hyperscaler's business model," said Josh Gilbert, lead APAC analyst at eToro. "Microsoft, Alphabet and Amazon can point their data centre dollars at cloud businesses that sell compute straight back out the door, but Meta doesn't have the same outlet, so every dollar of build-out leans on the ads business."
Zuckerberg painted a vision of AI-powered personal assistants used by billions and business agents handling customer service, sales, and marketing. But he offered few specifics on how those services would generate returns, leaving analysts frustrated. Bernstein analyst Mark Shmulik likened the earnings call to "a good old-fashioned brainstorming session."
The spending echoes Meta's costly metaverse pivot, which racked up tens of billions in losses without becoming a major revenue driver. The free cash flow slump was the steepest since late 2022, when metaverse bets drew similar scrutiny. Yet Meta raised its capital spending forecast by $5 billion to a range of $130 billion to $145 billion, mirroring Alphabet's recent $15 billion increase that triggered a 7% stock drop.
Scarce Chips, Vague Returns #
CFO Susan Li argued the industry had "underbuilt historically" for AI demand, making existing capacity "extremely valuable" for the foreseeable future. She said Meta could generate returns through products, enterprise services, and compute sales, but declined to specify where the greatest profits would lie.
This ambiguity unsettles investors. Microsoft, by contrast, showed how early AI bets pay off: its Azure cloud and Copilot assistant beat growth expectations, sending its stock up 8% despite a 23% free cash flow decline. Meta has no such enterprise pipeline.
Zuckerberg noted Meta was buying third-party capacity while also selling its own, a seeming contradiction. He explained it as investing ahead of demand: "There is a lead time where we're investing in building out these data centers now. They come online at some point in the future. You obviously are not getting value out of them until they're online."
The compute conundrum reflects a broader industry challenge. Research on AI infrastructure scaling shows that demand for training and inference compute is doubling every few months, forcing companies to overbuild or risk falling behind. Meta's bet is that owning scarce compute will become a competitive moat, even if the path to monetization remains hazy.
Ads Alone Can't Carry the Load #
Meta's core advertising business remains robust, but the AI spending spree is testing investor patience. The company's free cash flow of $784 million is a fraction of what's needed to sustain operations and growth, let alone fund ambitious AI projects.
Zuckerberg acknowledged the tension: "We believe that there will continue to be a significantly higher margin on selling intelligence rather than selling compute directly, but we think that there's a big opportunity obviously to sell compute as well." The statement suggests a dual-track strategy, but without a clear timeline or revenue split, it reads more like hedging.
Meta's predicament highlights a structural disadvantage. Cloud providers can amortize AI infrastructure costs across millions of enterprise customers, while Meta must extract value primarily from its consumer platforms. Meta's open-source Llama models could attract developers, but monetizing them remains an unsolved puzzle.
As the AI arms race intensifies, Meta's compute conundrum may force a strategic reckoning. Either it finds a way to turn AI into a revenue stream beyond ads, or it risks repeating the metaverse's financial hangover. For now, investors are left parsing vague promises while watching cash reserves dwindle.