Meta’s shares jumped after report that it plans to commercialize excess AI compute via a new cloud business

Meta Platforms shares rose sharply on July 1 after a report said founder and CEO Mark Zuckerberg is building a cloud-style business for customers, using excess artificial-intelligence compute that Meta currently runs for its own workloads. The move highlights how quickly the AI compute race is turning from pure internal investment into potential third-party services for the broader market.

The July 1 jump was widely reported as a 9% move to about $612.91, according to market coverage that tracked the day’s trading. The following day, shares reportedly fell about 5%, illustrating how investors are weighing the promise of faster monetization against the financial risk that comes with building and operating massive data-center capacity.

The underlying catalyst, described in the market commentary, is that Meta could sell surplus AI processing capacity to outside customers. In that framing, Meta would be creating a more hyperscaler-like revenue stream, one that could sit alongside its existing advertising and business messaging products by monetizing compute itself.

The market coverage also described the proposed venture by name, saying it would be called “Meta Compute” and would pitch to the same enterprise and developer demand that Amazon Web Services, Microsoft Azure, and Alphabet’s Google Cloud target. If Meta can deliver capacity as a service, the company would be positioning its AI infrastructure not just as a cost center, but as an offering that can potentially generate revenue from others’ AI workloads.

Investors appear to be focused on capital intensity and payback. The coverage cited Meta’s capital expenditures (capex) rising 84% year over year in 2025 to $72.2 billion, with projections for 2026 capex totaling between $125 billion and $145 billion. Those numbers, if accurate, would make Meta a more aggressively capital-intensive player, and they set a high bar for how quickly the company can earn meaningful returns on its spending.

The same commentary referenced Zuckerberg previously indicating that the company has options if it ends up building more capacity than it can absorb internally. Selling excess compute to outside customers is presented as one such route, but the post’s framing also implies that the market is uncertain about how scalable and profitable that business could become.

For Microsoft, the immediate relevance is competitive, not operational. Microsoft is already a dominant cloud provider for AI workloads, but Meta’s reported attempt to package compute resources for external customers would raise the prospect of a new challenger that can leverage its own scale and demand indicates from ongoing AI deployments. The real-world effect on Microsoft’s business would depend on execution, including whether Meta can offer compute at competitive price-performance levels and win reliable enterprise and developer customers.

Still, key details are missing from the market write-up itself. The report does not lay out product specifications, pricing, customer contracts, timeline for launch, or how “Meta Compute” would integrate with Meta’s existing AI stack and data-center strategy. Without those specifics, investors may be left to react to the idea of an additional compute revenue stream rather than a fully detailed, revenue-bearing plan. Watch for whether Meta later confirms the concept, provides timing and terms, or offers disclosures that connect capex spending directly to external commercialization.