Meta Platforms is building a full cloud infrastructure business to sell its excess AI computing capacity to outside companies, Bloomberg News reported on July 1, 2026. The internal initiative known as Meta Compute was quietly created back in January and is now in active development. The plan has two tracks: one would let developers access AI models hosted directly on Meta’s own infrastructure, including its Muse Spark models, and pay for the computing power used to run them similar to how Amazon Web Services’ Bedrock product works. The second track would see Meta rent out raw GPU capacity from its data centres to third parties by the hour, which is exactly the model used by so-called neocloud companies like CoreWeave and Nebius. Meta declined to comment on the report. Reuters said it could not independently verify the Bloomberg story, and the plans are still in development with the strategy subject to change.
The move would put Meta in direct competition with the three biggest players in cloud computing – Amazon, Microsoft, and Alphabet and represent a major shift for a company that, until now, has earned almost all of its money from digital advertising. Meta’s stock surged more than 10% on the news, clawing back some of the ground it had lost after underperforming the S&P 500 by nearly 15% this year as of Tuesday.
“Meta stock is absolutely flying today. Up over +10% on reports the company is building a cloud infrastructure business to sell access to AI compute power. All those massive CapEx investments in data centers and GPUs might actually pay off big time.”~Mario Nawfal
Zuckerberg Has Been Signalling This Move for Months:
The Bloomberg report did not come out of nowhere. Mark Zuckerberg first raised the possibility of a cloud business during Meta’s Q3 2025 earnings call and brought it up again at the company’s annual shareholder meeting in May 2026, where he told investors the idea was “definitely on the table.” He revealed that companies approach Meta “almost every week” asking to buy access to its computing infrastructure or AI models at a premium to what Meta paid for them. He added that while Meta had not done this yet because it believed it had uses for the compute internally, if it reached a point where it had overbuilt, selling access would absolutely be an option.That language was aimed, not a guarantee. Bloomberg’s piece today shows that the concept has evolved much beyond a contingency: Meta Compute now has a name, an inside staff, and active planning.
Meta has committed to spending between $125 billion and $145 billion on capital expenditure in 2026 alone, including a 2,250-acre hyperscale campus in Louisiana called Hyperion and a one-gigawatt data centre under construction in the American Midwest. Big Tech firms collectively are expected to spend more than $700 billion on AI infrastructure this year, up from roughly $400 billion in 2025. The pressure on these companies to show returns on that investment has been growing steadily, and a cloud business that monetises idle capacity would go a long way toward answering investor concerns about overspending.
“$META BREAKING: Meta is now building a cloud business to sell excess AI compute, as per Bloomberg. If this is true…then it is exactly what the market has been waiting for. Justifies the heavy capex spend. Diversifies a line of business outside of advertising.”~amit
CoreWeave and Nebius Cratered as Meta’s Entry Rattles Neocloud Sector:
Although Meta shareholders cheered, the neocloud sector suffered a setback. CoreWeave’s shares sank about 14%, while Nebius fell 12% to 15% in a single session, showing how heavily the sector relies on Meta as a customer. CoreWeave has a $21 billion contract with Meta until December 2032, while Nebius has an agreement for up to $27 billion over five years. When those contracts were inked, neither business was in direct competition with Meta. If Meta begins selling its own surplus compute outside, it lessens its need to purchase capacity from them while also entering the same market they operate in.
According to Gil Luria, managing director at D.A. Davidson, adding Meta’s capacity to the market will have a greater influence on neoclouds than the big hyperscalers. Companies like CoreWeave and Nebius rely on Meta for growth, but Meta may no longer require them.” Bernstein analysts saw the development as “problematic” for CoreWeave, saying that hyperscaler competition had always been a danger for pure-play neocloud providers. The problem is structural, not simply about today’s pricing: CoreWeave’s $8.5 billion lending facility, which is backed by contracted GPU cash flows, might deteriorate if Meta’s long-term contracts are revised or not renewed.
“$META is reportedly developing a cloud business to sell access to excess AI compute, per Bloomberg. The internal initiative is called Meta Compute. The plans being considered: AI model access hosted on Meta infrastructure, similar to AWS Bedrock. Raw AI compute capacity.”~Wall St Engine
Meta Is Following SpaceX’s Playbook And the Market Is Watching:
Meta is not the first major AI spender to look at turning infrastructure into a revenue stream. Elon Musk’s SpaceX, through its xAI subsidiary, began selling excess computing capacity earlier this year, renting out all of its Colossus 1 data centre capacity to Anthropic and striking further deals with Google and Reflection AI. That strategy could help xAI generate more than $50 billion in revenue by 2028, according to Bloomberg Intelligence estimates. Meta, with a capex programme several times larger, is now looking at a similar playbook.
Meta’s position is remarkable because of the enormous scale it brings to the situation. It is not entering the cloud market as a startup or a niche GPU provider; rather, it is arriving with infrastructure that rivals or exceeds that of most specialized cloud firms, while avoiding the GPU-collateralized debt most neoclouds use to finance their gear. If the cloud business takes shape as expected, it could significantly reduce Meta’s reliance on advertising, provide investors with a stronger argument for the company’s massive AI spending, and disrupt competitive dynamics in both the cloud and neocloud sectors in ways that are still difficult to fully understand.
“Meta Platforms is building a cloud business to sell excess AI computing capacity, Bloomberg News reported, as tech giants seek returns on costly AI investments amid worries about overspending. Meta shares were up more than 7% in early trading.”~Reuters
