Meta Platforms plans to spend hundreds of billions of dollars to build several large AI data centers, CEO Mark Zuckerberg said on Monday, as the company increases its push into artificial intelligence.
In a post on Threads, Zuckerberg said Meta’s first multi-gigawatt data center, named Prometheus, is expected to go live in 2026. Another facility, Hyperion, will scale up to 5 gigawatts over time.
He said Meta is also building several more large “titan clusters,” with one covering an area close to the size of Manhattan.
The company reorganized its AI efforts last month into a new unit called Superintelligence Labs, following challenges with its open-source Llama 4 model and the departure of key staff. The new division is expected to support Meta’s AI products, including its Meta AI app, image-to-video ad tools, and smart glasses.
Zuckerberg said the company’s advertising business provides the financial strength needed to support this investment. “We have the capital from our business to do this,” he said.
Meta generated nearly $165 billion in revenue last year. Its stock rose about 1% on Monday and is up more than 20% this year.
Zuckerberg has also been involved in a recent push to hire talent for the AI division. Superintelligence Labs will be led by Alexandr Wang, formerly of Scale AI, and Nat Friedman, former CEO of GitHub.
Meta invested $14.3 billion in Scale before appointing Wang.
In April, Meta raised its 2025 capital spending forecast to between $64 billion and $72 billion to support its AI expansion and remain competitive with companies like OpenAI and Google.
Zuckerberg also highlighted a report from SemiAnalysis that said Meta may become the first AI lab to bring a gigawatt-plus supercluster online.
