Meta Builds Cloud Business to Sell Excess AI

Meta Is Building a Cloud Business to Sell Excess AI Compute

We're covering this story as it develops — the plans are still in flux, but the market reaction was immediate.

Meta Platforms is building a cloud business to sell excess AI computing capacity. Bloomberg reported this on July 1, 2026, citing people familiar with the matter. The initiative is called Meta Compute.

The service would let developers access AI models hosted on Meta's infrastructure. They would pay for the computing power needed to run them. Meta is also considering selling raw AI computing capacity — similar to what neocloud companies currently offer.

The plans are still in development. Bloomberg's sources noted the strategy could still change.

What Is Meta Compute?

Meta Compute would work like Amazon Web Services' Bedrock product. Bedrock lets developers access AI models from multiple companies in one place. Meta's version would do the same, using models hosted on its own infrastructure.

One of those models would be Muse Spark. Meta unveiled Muse Spark in April 2026. It was the first model from a high-cost AI research team Meta assembled. As of the Bloomberg report, Meta had not released it to developers. A Wall Street Journal report said last month there is no scheduled launch date.

Meta CEO Mark Zuckerberg signaled this move was coming. At the company's annual shareholder meeting on May 27, he said entering cloud computing was "definitely on the table." He added that companies were approaching Meta "almost every week" to buy access to its AI models or spare computing power.

You might also like

Palantir-NVIDIA Nemotron Engine for US Gov AI

Tesla Optimus Line Starts at Fremont in July 2026

Microsoft Cuts ~9,000 Jobs in July 2026 Round

Neuralink Skips Dura Cut in New Brain Implant

Why Is Meta Spending So Much on AI?

Meta is projected to spend as much as $145 billion on AI infrastructure in 2026. That is a large share of Big Tech's total outlay of more than $700 billion on AI this year.

That level of spending creates pressure to show returns. Selling excess compute is one way to earn revenue from infrastructure that would otherwise sit idle.

Gil Luria, managing director at D.A. Davidson, compared the situation to SpaceX: "This is very similar to the situation SpaceX has found itself in, which led it to sell compute capacity as well." SpaceX recently struck deals to rent out its Memphis data center access to Anthropic and Google.

This dynamic — massive infrastructure investment seeking a return — mirrors what's happening across the AI sector. OpenAI's revenue targets and IPO valuation pressure reflect the same underlying tension.

How Did Markets React?

The market reaction on July 1 was sharp and split:

Meta shares had already fallen nearly 15% year-to-date as of Tuesday. The cloud report reversed that pressure in a single session.

Neocloud companies took the hardest hit. Luria explained why: "The impact of adding Meta's capacity to the market is more likely to be on neoclouds than the big hyperscalers. Those companies like CoreWeave and Nebius rely on Meta for their growth and Meta may not need them anymore."

What Rosenblatt Says About CoreWeave

Rosenblatt maintained its Buy rating and $250 price target on CoreWeave after the sell-off, according to Investing.com. The firm called the stock weakness a buying opportunity.

Rosenblatt's checks showed no change in hyperscale demand for GPU compute. GPU shortages remain the norm across the industry.

The firm also flagged a key legal detail: Meta does not have the right to resell capacity it has leased from CoreWeave through 2032 to third parties. That limits how quickly Meta could bring leased compute to market as a product.

Key points from Rosenblatt's analysis:

• No change in hyperscale posture toward GPU procurement
• GPU shortages remain the norm industry-wide
• Meta cannot resell CoreWeave-leased capacity to third parties through 2032
• The sell-off represents a buying opportunity, per Rosenblatt

Broader Context: AI Infrastructure Race

Meta's move fits a wider pattern. Companies that built massive AI infrastructure are now looking for ways to monetize it. The AI chip deals being pursued across the industry reflect the same logic — scale first, then find the revenue model.

Analysts also noted the cloud pivot raised doubts about Meta's AI lab ambitions. Meta has backed its AI push with billions in investment, including a talent war last year. But competing with AWS, Azure, and Google Cloud is a different challenge from building frontier AI models.

The Trump administration has separately asked Meta to submit Muse Spark for voluntary government review. Other major U.S. AI developers have already taken that step.

Meta declined to comment. Reuters said it could not independently verify the Bloomberg report.