Mark Zuckerberg, CEO of Meta, has announced plans to invest hundreds of billions of dollars in the development of massive AI-focused data centers across the United States.
The first of these large-scale facilities, named Prometheus, is scheduled to become operational by 2026 and will be located in New Albany, Ohio. According to Zuckerberg, one of the sites will span an area nearly as large as Manhattan—about 59 square kilometers (22.8 square miles).
Meta has been significantly increasing its investment in artificial intelligence, particularly in the development of what it refers to as “superintelligence”—a form of AI designed to exceed human cognitive abilities.
In a post on Threads, Zuckerberg shared that the company is constructing multiple high-capacity data clusters, including a site named Hyperion, which will be based in Louisiana. Hyperion is expected to scale up to five gigawatts of computing power over the next few years and be fully operational by 2030.
“We’re developing several more massive clusters. Just one of them takes up an area comparable to a large portion of Manhattan,” Zuckerberg said. He added that the facilities were given mythological names to reflect their scale and technological significance.
Meta, whose core business is still driven by digital advertising, brought in over $160 billion in revenue in 2024. The company’s expansion into AI is seen as part of its broader strategy to remain competitive in the evolving tech landscape.
Industry experts view this move as a bold attempt to become a leading force in AI. Karl Freund, principal analyst at Cambrian AI Research, told the BBC, “Zuckerberg clearly plans to buy his way to the top of the AI world. The engineers he’s bringing on board will be working with some of the most advanced AI hardware available.”
Following the announcement, Meta’s stock rose by 1%, with shares up more than 20% so far this year, according to Reuters.
Globally, there are over 10,000 data centers, many of them in the United States, with the UK and Germany also hosting significant numbers. However, AI-powered data centers come with high environmental costs. A recent study suggests these facilities could use up to 1.7 trillion gallons of water worldwide by 2027. Even a single AI query—such as one to ChatGPT—can consume as much water as a small bottle purchased at a local store.
