When a social media titan starts selling off its surplus computing power, markets pay attention — and with good reason. On July 1, Meta Platforms (META) announced the launch of “Meta Compute,” a cloud service that rents out excess AI infrastructure, putting it on a collision course with Amazon Web Services, Microsoft Azure, and Google Cloud. The stock rocketed 9% to $612.91 in a single session, only to surrender nearly 5% the very next day.
With Meta trading at a steep discount to big-tech peers, this abrupt strategic shift has ignited a charged debate: is the seemingly cheap stock a bottomless value trap, or a golden opportunity hiding in plain sight?
Skeptics argue that selling compute capacity amounts to a tacit confession that Meta overinvested in AI infrastructure. The numbers are staggering: capital expenditures surged 84% year over year to $72.2 billion in 2025, with projections of $125 billion to $145 billion for 2026 — a dramatic transformation from an asset-light platform into a capital-hungry enterprise. Yet despite pouring hundreds of billions into open-source AI models that remain largely free, the company has no clear internal monetization path. Turning landlord, critics say, looks less like a savvy move and more like an early warning signal that the AI boom is on shaky ground. If even the most aggressive builder cannot absorb its own capacity, how solid is industry-wide demand?
The specter of the metaverse disaster looms large. Not long ago, Mark Zuckerberg rebranded Facebook to Meta and burned billions on a virtual world that never materialized. Should history repeat, not only would the AI investments struggle to earn a return, but the crushing maintenance costs of vast data centers would erode profits for years. That’s why bears insist that until Meta proves AI can reliably generate cash, the stock deserves its deep valuation haircut.
Optimists see a very different picture. With AI demand exploding, raw computing power has become a scarce resource. Reports that Alphabet is paying SpaceX a staggering $920 million per month for AI compute underscore just how tight supply has become. By entering the market now, Meta is converting a sunk cost into a recurring revenue stream at a moment when demand far outstrips supply. Far from a sign of weakness, acting as a “picks-and-shovels” supplier is the most rational commercial decision the company could make.
Crucially, Meta’s foundation remains rock-solid. Advertising revenue — still the lion’s share of the business — jumped 33% year over year in the latest quarter, showing the cash machine is running at full throttle. This gives Meta Compute ample financial cover to find its footing. And even if AI ambitions ultimately disappoint, the underlying infrastructure can be repurposed into a mature public cloud offering, a trillion-dollar playbook already validated by Amazon and Microsoft. The company still holds plenty of cards.
At its core, Meta’s predicament is a referendum on capital discipline. Runaway spending paired with uncertain monetization has already sent the stock down 26% from its all-time high last August, fully pricing in those anxieties.
Yet the decision to sell compute marks a meaningful inflection. It signals that management is no longer building in a vacuum but is instead confronting reality and choosing a shorter, faster route to returns — a potential escape from the vicious cycle of heavy investment and opaque payoffs. For long-term investors who believe the AI revolution is just beginning, Meta’s newfound strategic flexibility, massive computing assets, and deeply discounted valuation could indeed form a golden entry point. Whether the company can hold its own against entrenched cloud titans, however, will ultimately determine if this pivot marks the start of a sustained rebound or simply another chapter of value destruction.
