What happened
Meta's stock jumped more than 6% on 1 July 2026 after reports that the company is building a cloud business to sell its excess AI computing power. An internal organisation called Meta Compute sits at the centre of these efforts, led by Santosh Janardhan, Meta's head of infrastructure; Daniel Gross, a leader inside the Meta Superintelligence Labs AI unit; and Meta President Dina Powell McCormick.
The surplus exists because Meta is spending roughly $125 to $145 billion on AI infrastructure in 2026, up from about $72 billion last year, likely more than its own products need. CEO Mark Zuckerberg signalled openness to such a move at the company's shareholder meeting in May, saying the idea of selling computing access was something the company had been actively fielding and that it was "definitely on the table."
Meta plans to provide API access to its AI products, including the Muse Spark model that launched in April 2026, and will offer raw GPU rentals, allowing developers and companies to use Meta's computing hardware to run their own AI workloads. The plans would put Meta in direct competition with Amazon Web Services, Microsoft Azure, and Google Cloud.
Why it matters for social marketers
Meta has historically been a single-product company. In its earnings report last month, Meta said that nearly 98% of its $56.3 billion in first-quarter revenue came from advertising. This cloud move changes how marketers should read Meta's infrastructure decisions.
Every dollar Meta spends on AI capacity is no longer purely defensive or speculative. It can now be recouped through external sales, meaning Meta's appetite for infrastructure risk shifts. That has two implications for anyone running social campaigns.
First, it reduces the pressure on the advertising business to justify Meta's infrastructure costs alone. Investors have questioned whether Meta can generate returns from its AI buildout. Selling compute capacity gives the company a second revenue stream and changes the return-on-investment calculus for every GPU rack they install. Analysts believe that selling excess compute capacity could reduce earnings risk for Meta by providing higher near-term revenue compared to waiting for other AI products.
Second, it repositions Meta's AI infrastructure as a competitive asset, not just a cost centre. Mark Zuckerberg framed infrastructure as a strategic advantage, writing that "Meta is planning to build tens of gigawatts this decade and hundreds of gigawatts or more over time" and that "how we engineer, invest and partner to build this infrastructure will become a strategic advantage."
For social media professionals, that means the compute underpinning Meta's ad targeting, creative generation tools, and Advantage+ automation is now being built at a scale and pace that would be unsustainable if ads were the only business case. The infrastructure becomes more robust because it serves two masters.
What it changes on the ground
Meta's advertising platform has become heavily reliant on AI-driven automation. According to Meta's reporting, Advantage+ campaigns are delivering 22% higher ROAS on average compared to manually managed campaigns. Advertisers who turned on Advantage+ creative's AI-driven targeting features experienced a 22% increase in ad ROI, and businesses using image generation saw a 7% increase in conversion rates.
The cloud business de-risks that AI investment trajectory. Meta can justify continued spend on inference capacity, model development, and GPU clusters even when ad revenue fluctuates, because those same resources can be monetised externally. That should translate into faster rollout of AI tools for advertisers, more aggressive feature launches, and less pullback during softer quarters.
It also signals confidence. Companies don't productise and sell infrastructure they think they'll need themselves in six months. By opening its capacity to external customers, Meta is effectively stating that its own AI roadmap for ads, content recommendation, and creator tools won't consume everything it's building. Whether that confidence is justified remains to be seen, but the market is clearly betting it is.
The AWS comparison everyone's making
Some analysts see it as a natural evolution: just as Amazon turned its internal network expertise into AWS, Meta can commercialise its AI prowess. The parallel is instructive but not perfect. AWS succeeded because Amazon faced a genuine internal capacity problem and had to build infrastructure that could handle spiky retail demand. That forced discipline produced a product others wanted.
Meta's situation is different. The company is spending around $145 billion this year on AI infrastructure, roughly double what it spent in 2025. The question isn't whether Meta built something impressive. It's whether enterprises will trust a social media company to run their mission-critical workloads, and whether Meta has the sales operation, compliance stack, and enterprise support culture to compete with vendors who've been doing this for 15 years.
AWS, Microsoft and Google have spent years building trusted cloud platforms with global data centres, enterprise sales teams, security systems and compliance tools, and customers tend to be cautious about moving critical workloads from one cloud provider to another. Meta has scale, but it doesn't yet have trust in the enterprise cloud market, and earning that takes time.
What to watch
This is still a reported plan, not a confirmed product. A Meta spokesperson declined to comment, and the company's plans are still in development, meaning it's possible the strategy could change. But the market's reaction suggests investors believe it's real.
Meta shares rose 8.8% on 1 July. Shares of CoreWeave and Nebius fell 10.8% and 12.4%, respectively, following the report. The neocloud providers that currently supply Meta with additional capacity are now facing a new competitor: their own customer.
For social media marketers, the commercial signal is clear. Meta's infrastructure strategy is no longer just about making better ads. It's about becoming a diversified technology company where advertising sits alongside cloud services. That doesn't diminish the ad business. It just means the platform you're buying media on is now playing a different game at the infrastructure layer, one with implications for how aggressively Meta invests, how it prices risk, and where it sees its competitive advantage in five years.
The specifics of what Meta will sell, to whom, and at what price remain unclear. But the direction is set. The company that spent a decade convincing marketers it was more than just Facebook is now convincing investors it's more than just ads. Whether that makes it a better or worse partner for marketers depends on whether the cloud business strengthens Meta's infrastructure position or distracts from the 98% of revenue that still comes from the people who pay to reach its 3.98 billion users.