Meta shares jump over 8% on plans to sell AI compute, shaking up the cloud market

On July 1, shares of Meta jumped 8.81% to finish at an all-time high of $796.25 after the release of the news about the company’s plans to use its surplus AI computing power in its data centers for commercial purposes. This increase helped make Meta one of the best-performing stocks on Nasdaq, although the rest of the market fell short. The S&P 500 index closed down 0.22% at 7,483.23.

Technology stocks suffered as well, with the Nasdaq Composite index losing 0.66% (to 26,040.03) and the PHLX Semiconductor index falling 6.27% due to a large sell-off in stocks of AI and semiconductor companies.

The decision to rent excess AI computing power is likely to put Facebook’s and Instagram’s parent now in competition with major public cloud vendors [Microsoft, Amazon Web Services (AWS) and Alphabet] and newer companies such as CoreWeave, who provide GPU rental services to customers who develop AI applications and processes, because Meta’s use of excess computing capacity may limit the use of these companies for rental services.

Reuters stated that investors viewed this decision as a long-term source of return for Meta, given their rapidly growing and expensive AI-related infrastructure investments, and as a means of reducing revenue dependence on advertising.

Gil Luria, managing director of D.A. Davidson, believes that most of the impact in the market will be felt by AI-focused data center operator firms instead of large-scale cloud service providers, noting that “adding Meta’s capacity to the market is more likely to be on neoclouds than the big hyperscalers.” He added that companies such as CoreWeave and Nebius rely heavily on Meta for growth and “Meta may not need them anymore.”

Meta looks to monetize excess AI compute

Reports suggest that Meta may create a commercially available way for customers to access spare compute capacity from its exponential growth in its AI data center network, as well as provide developers the ability to run AI models on its infrastructure.

The model reported to be under consideration appears to be similar in design to AWS’ Bedrock platform, where users pay to access hosted foundational models via APIs, with the alternative being considered is similar to CoreWeave’s rental of raw GPU compute capacity. Current development of such plans continues, and no plans have been made public for a formal launch or commercialization at this time.

Considering that Meta has historically created AI infrastructure solely for the development of internally-facing products, this shift represents an important commercial evolution as well. In reference to this shift, during the most recent shareholder meeting in May 27, 2026, CEO Mark Zuckerberg made note of an increasing volume of interest being expressed from companies outside of Meta in its AI infrastructure when he said:

“It’s definitely on the table. Almost every week different companies come to us from the outside asking us to both stand up an API service or asking if we have compute that they could buy from us at some premium to what we’ve bought it at.”

Zuckerberg added that Meta had not pursued the opportunity because it still expected to use the capacity internally, but said selling excess compute would become an option if the company ultimately overbuilt its infrastructure.

The announcement comes at a time when investors are increasingly questioning the return on massive AI capital expenditure across big tech. Meta, Microsoft, Alphabet, and Amazon are collectively expected to spend roughly $725 billion on AI infrastructure in 2026.

Meta alone expects capital expenditures of up to $145 billion, one of the largest AI investment programs in the technology sector. Monetizing excess compute could provide investors with a clearer path toward generating revenue from those investments.

Although Meta has yet to announce pricing or packaging of its proposed cloud services, it is entering an already-established market. AWS usually charges on a pay-as-you-go basis for GPU instances or API usage through Amazon Bedrock, while charging separately for separate instances of model inference and compute resources based on the individual hardware and specific model chosen.

However, CoreWeave focuses exclusively on the rental of dedicated high-performance NVIDIA GPU clusters for AI training/inference via only enterprise contracts and reserved capacity contracts.

As such, industry analysts will watch if Meta competes on price, uses its own infrastructure to undercut its competition pricing-wise, or offers access to its proprietary AI models as part of its compute service bundle offering.

Investor sentiment is still fragile

While July 1’s rally partially offsets a challenging period for the stock, Meta is still down almost 8% since the start of the year, reflecting continuing doubts from investors regarding whether significant AI investments will result in reliable growth of revenues.

Selling off excess computing resources will address this issue directly; it will enable Meta to turn excess data centre capacity into income-generating assets rather than fixed costs. It will also enable Meta’s business to diversify to include more than digital advertising by developing an enterprise infrastructure business similar to Amazon, Microsoft, and Google’s successful ones, which have become key engines of profit for those three companies.

Moreover, Meta has decreased its workforce and greatly increased its level of investment in AI. Consequently, the financial results of its infrastructure business strategy became even more critical to investors assessing Meta’s long-term profitability.

Meta must clarify pricing, models and launch timing

It is still uncertain whether or at what scale Meta intends to launch their cloud offering. The project currently exists as a work-in-progress rather than as a finished product, and to this point, Meta has not publicly disclosed any information regarding the service.

Investors will want to keep a close eye out for additional information on Meta’s AI cloud offering, particularly with regard to the pricing structure compared to key competitors like AWS, Google Cloud, Azure, and CoreWeave. Other crucial things to look out for include:

• whether or not Meta will provide hosted AI models, raw GPU compute, or both at launch
• whether or not any of its flagship models will be accessible through APIs
• when to expect news about any enterprise partnerships or early adopters
• and whether or not this new revenue source has enough potential to offset the company’s rapidly increasing costs associated with building an AI infrastructure.

If launched, this would be the first significant effort by Meta to create a direct revenue stream from their AI infrastructure, which could fundamentally change the way hyperscale cloud providers compete with one another as well as how the fast-growing AI compute market evolves.

Don’t just read crypto news. Understand it. Subscribe to our newsletter. It's free.