Meta Expands Computing Power to Advance AI

Meta is launching several new computing clusters, each with a capacity of 1 GW. The first cluster, named Prometheus, is expected to go live in 2026, while another, Hyperion, will scale up to 5 GW in the coming years. According to Mark Zuckerberg, one of these clusters will occupy an area comparable to Manhattan.

In his post, Zuckerberg stated that Meta intends to invest hundreds of billions of dollars into computing infrastructure to create superintelligence. He emphasized that the company has sufficient capital to execute these plans independently, without external funding.

Zuckerberg cited an analysis from SemiAnalysis, which indicates that Meta is ahead of its competitors and could be the first company to launch a cluster with a capacity exceeding 1 GW. Meta is currently building clusters even in specially designed tents to expedite the deployment of its computing infrastructure. This initiative is part of the transition from "AI incrementalism" to full-scale racing for AGI (Artificial General Intelligence).

The company is also assembling the "most elite" development team: it has already recruited top talents from OpenAI and Apple, and invested in Scale AI. The CEO, Alexander Wang, is leading the development team.