Meta is preparing to launch a cloud infrastructure business that would resell AI compute and hosted models, putting it in direct competition with Amazon Web Services, Google Cloud, and Microsoft Azure. The new initiative, reportedly dubbed Meta Compute, follows $182.9 billion in AI infrastructure commitments the company had booked by the end of the first quarter. The move mirrors a similar pivot by SpaceX, which began selling excess compute from its xAI-linked data centers earlier this year.
The business would sell raw compute capacity in the style of CoreWeave and also host access to specific AI models, including Meta's recently launched closed-weight model Muse Spark. That dual approach — infrastructure plus model-hosting — is the same combination that anchors AWS's generative AI offering. Meta declined to comment.
Meta Compute is being run by head of infrastructure Santosh Janardhan, Meta Superintelligence Labs leader Daniel Gross, and president Dina Powell McCormick. The leadership mix signals that Meta is treating the cloud business as a core strategic bet rather than an infrastructure side project, pulling in both its infra chief and the executive running its superintelligence effort.
Key facts
- 01Meta had committed $182.9 billion to AI infrastructure by the end of Q1, including data center projects in Louisiana and Ohio.
- 02The new business line, Meta Compute, is led by infrastructure chief Santosh Janardhan, Meta Superintelligence Labs' Daniel Gross, and president Dina Powell McCormick.
- 03Meta plans to sell raw compute capacity in the style of CoreWeave and hosted models including its closed-weight Muse Spark, per Bloomberg.
- 04The Ohio data center, which Zuckerberg said would be the size of Manhattan, is expected to come online this year.
- 05SpaceX moved first via xAI, signing an early-May deal with Anthropic to buy out all compute at Colossus 1, followed by leases with Google and Reflection AI.
The commercial logic is straightforward. Meta doesn't break out revenue from Meta AI or its Llama open-weight model family in earnings reports, and executives have leaned on internal corporate use cases when discussing AI's contribution to the business. That leaves Meta's AI spend without a matching standalone revenue line — a gap that reselling compute could partially close.
SpaceX set the template. In early May, it signed a deal with Anthropic to buy out all compute capacity at its Colossus 1 data center, then followed with similar leases to Google and Reflection AI. The pattern turned excess capacity into contracted revenue almost immediately, and it demonstrated that frontier AI labs will pay for guaranteed access rather than wait in line at the hyperscalers.
Zuckerberg signaled the direction in May, saying a Meta cloud computing business was on the table as a way to earn a return on the company's spending toward AI superintelligence.
“definitely on the table”— Mark Zuckerberg, Meta CEO
The Ohio build is central to the capacity story. Zuckerberg has described the site as the size of Manhattan, and it is expected to come online this year. Combined with the ongoing Louisiana project, Meta will have compute footprint well beyond what its own products currently consume — the exact overhang that a wholesale cloud business is designed to absorb.
The strategy carries risk. Some skeptics have warned that the AI infrastructure race is a bubble leaning heavily on rapidly depreciating chips, and others have questioned whether AI companies can generate enough end-user revenue to justify trillion-dollar buildouts. If demand for training and inference compute softens before Meta's data centers are contracted out, the same facilities that look like a revenue opportunity today become a stranded-asset problem tomorrow.
The signal Meta and SpaceX are jointly sending is that the AI race may be won at the infrastructure layer, not the model layer. Neither company is a traditional cloud provider, and neither has a hit consumer AI product on the scale of ChatGPT. What they have is capacity — and in a market where OpenAI, Anthropic, and other labs are compute-constrained, capacity is the scarce good. Expect the hyperscalers to respond on price and contract terms as two well-capitalized competitors enter their market from an unexpected direction.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
