July 1, 2026, (Inside AI) — Meta Platforms is developing a cloud computing service to sell surplus artificial intelligence infrastructure, according to a Bloomberg News report citing unnamed sources. The initiative would mark a significant pivot for the social media giant, potentially creating a new revenue stream beyond advertising.
The plan involves offering external customers access to AI models hosted on Meta’s existing data center infrastructure, mirroring Amazon Web Services’ Bedrock platform. This move would position Meta directly against established cloud leaders Amazon, Microsoft, and Alphabet.
Meta’s stock surged nearly 6% in premarket trading following the report. The company has not officially commented, and Reuters could not independently verify the details.
The strategic shift comes as Big Tech collectively pours unprecedented sums into AI infrastructure. Industry spending is projected to exceed $700 billion this year, a sharp rise from approximately $400 billion in 2025. Meta’s own capital expenditures have ballooned, driven by CEO Mark Zuckerberg’s aggressive bet on AI.
At the company’s annual shareholder meeting in May, Zuckerberg openly acknowledged the possibility. He stated that Meta is frequently approached by other firms seeking to license its AI models or purchase spare computing capacity at a premium. He framed the cloud option as a hedge against overbuilding.
Zuckerberg’s exact words were:
“It’s definitely on the table. Almost every week, other companies approached Meta asking it to sell them access to its AI models the way cloud providers do or looking to buy its spare computing capacity at a premium.”
He added:
“We haven’t done that yet, because we think that we have a use for the compute. But obviously, if we get to a point where we feel that we have overbuilt, then that is an option that we have, and that is partially what gives us confidence in investing in building this out.”
Meta’s potential entry into cloud services is not without precedent. The company has long operated massive data centers to support its family of apps. However, selling raw compute or AI model access would be a departure from its consumer-focused history. It echoes Amazon’s evolution from e-commerce operator to cloud powerhouse with AWS.
The reported model resembles a managed AI service rather than bare-metal infrastructure. By hosting models like Llama on its own servers and offering them via API, Meta could compete in the fast-growing “AI-as-a-service” market. This approach lowers the barrier for enterprises that lack the expertise or capital to run large models themselves.
Yet the move raises questions. Meta’s AI infrastructure is optimized for its own workloads—recommendation systems, content moderation, and generative AI features. Repurposing that capacity for external clients introduces technical and operational complexity. Security, isolation, and service-level guarantees would need to match hyperscaler standards.
Financially, the cloud business could diversify Meta’s revenue, which remains heavily dependent on advertising. In 2025, ads accounted for over 98% of Meta’s income. A successful cloud unit could provide a buffer against ad market volatility and regulatory pressures.
Competitors are unlikely to stand still. AWS, Microsoft Azure, and Google Cloud have deep enterprise relationships and vast service catalogs. They also offer their own frontier AI models. Meta would need to differentiate on price, performance, or unique model access—perhaps by granting early or exclusive use of future Llama versions.
The timing aligns with broader industry trends. Enterprises are increasingly seeking alternatives to dominant cloud providers to avoid vendor lock-in. Meta’s brand and scale could attract customers looking for a neutral, AI-focused platform. However, trust in Meta’s data practices may be a hurdle, given its history of privacy controversies.
Zuckerberg’s comments suggest the cloud plan is a contingency, not a committed launch. The company still believes it can productively use all its compute internally. But the sheer scale of investment creates a safety net: if internal demand lags, external sales can absorb the excess.
This strategy mirrors how other tech giants manage capacity. Google and Microsoft also use their clouds to monetize surplus infrastructure. For Meta, the difference is starting from scratch in a mature market. It would need to build sales, support, and partner ecosystems rapidly.
The reported exploration signals a maturation of Meta’s AI ambitions. From open-source model releases to potential cloud services, the company is seeking multiple paths to monetize its AI investments. Whether it can execute without distracting from its core business remains an open question.
