Meta is building a cloud computing business to compete with Amazon, Microsoft, and Google while monetizing its massive investment in AI data centers. The move could help offset Meta's steep AI spending by selling cloud services and its own AI models to customers, while also addressing investor concerns that AI costs will outpace returns.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Meta is developing a cloud infrastructure business that would compete directly with Amazon Web Services, Microsoft Azure, and Google Cloud, according to Bloomberg. The company has been investing heavily in data centers and would use the service to sell excess computing capacity and its own AI models.
Why it matters
Meta has faced investor concerns about its aggressive AI spending—the company raised its 2025 capex forecast to $125 billion（約20兆円） to $145 billion（約23兆円）. A cloud business could offset those costs by creating a revenue stream from excess capacity, while also giving Meta the competitive advantage that other cloud providers already enjoy by bundling AI models with infrastructure services.
What to watch
A Bank of America analyst found Meta's AI infrastructure costs are significantly lower than Wall Street expected—the company estimates roughly $22 billion（約3.5兆円） per gigawatt instead of the $45 billion（約7.2兆円） originally forecast. Meta is also developing custom AI chips called MTIA with Broadcom, with the first chip (Iris) scheduled to begin production in September.
Meta's cloud infrastructure push represents a strategic response to its mounting AI capital costs, which have drawn skepticism from investors worried the company is spending too much before proving returns. By creating an outlet to monetize excess data-center capacity and sell its own AI models—much like Amazon Bedrock, Google Cloud Vertex AI, and Azure AI Foundry already do—Meta would not only generate new revenue but also level a competitive playing field that has so far favored established cloud giants. The company's discovery that its cost per gigawatt is roughly half Wall Street's estimate suggests the underlying economics of this cloud business may be more favorable than consensus expected, potentially making the capex investment more defensible to shareholders.
Parallel developments in custom chip design (the MTIA chips built with Broadcom) reinforce this strategy by reducing the operating costs of AI workloads once deployed. Combined with competitive pricing for its latest AI model, Muse Spark 1.1, Meta is positioning itself not just as a consumer AI company but as an infrastructure and services provider for developers and enterprises, a role traditionally dominated by the Big Three cloud vendors.
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
