
Meta is developing a cloud business to sell AI compute power and hosted models, aiming to recoup its $182.9 billion(約29兆円) commitment to data center infrastructure. The company has not generated material revenue from its own AI services, making compute sales a potential path to returns. The move mirrors SpaceX's strategy and reflects a broader belief that data center ownership, not model quality alone, may determine AI business winners.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Meta is developing a cloud infrastructure business to sell access to AI compute power and models, according to Bloomberg. The initiative, reportedly called Meta Compute, is led by infrastructure head Santosh Janardhan, Superintelligence Labs leader Daniel Gross, and president Dina Powell McCormick. Meta is considering selling both raw compute capacity and access to AI models, including its recently launched closed-weight model Muse Spark.
Why it matters
Meta has committed to spending $182.9 billion(約29兆円) on AI infrastructure in the coming years but has not seen significant revenue from its own AI models and services—it does not break out Meta AI or Llama revenue in earnings reports. By selling compute and model access, Meta could recover returns on its massive infrastructure investments, mimicking a business model that competitors like SpaceX (via xAI) have already begun pursuing.
What to watch
Meta's Ohio data center project, which is expected to come online this year, is described as the size of Manhattan. The move also signals a broader trend: some analysts suggest the real winners in the AI race may be those who own and operate data centers, not necessarily those providing the best consumer AI models and services—though skeptics warn that the AI infrastructure buildout may be creating a bubble dependent on rapidly depreciating chips.
No discussion yet for this article
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack