Meta’s historic investment in artificial intelligence

Meta is making massive AI investments: Plans to spend hundreds of billions

Meta CEO Mark Zuckerberg has announced plans to invest hundreds of billions of dollars in building some of the world’s largest AI superclusters.

Meta revealed it's building multi‑gigawatt data centers to support AI. On Monday, Zuckerberg confirmed these investments aim to create the largest AI clusters globally. The first supercluster, named “Prometheus,” is slated to go live in Ohio in 2026, with several other gigawatt‑scale clusters underway.

One project, “Hyperion,” is designed to reach a total compute power of 5 gigawatts over the next few years and will cover a substantial portion of Manhattan Island. In addition, many other mega‑clusters are planned.

These superclusters are crucial for training complex AI models, generative AI, computer vision, and robotics, providing large-scale compute resources.

Zuckerberg emphasized: “Meta Superintelligence Labs will be industry‑leading and offer the most compute power per researcher.” These investments are central to Meta’s long‑term AI strategy.

This announcement comes as Meta increases hires and investments in AI. In June, the company announced a $14 billion Scale AI deal. The newly formed Meta Superintelligence Labs brings together top AI researchers and engineers.

Heading the unit are former Scale AI CEO Alexandr Wang and former GitHub CEO Nat Friedman. Meta has made a $14.3 billion investment in Scale AI and is aggressively recruiting top talent.

This rapid expansion is driven by Zuckerberg’s dissatisfaction with Meta’s pace in AI. After limited interest in the launch of the Llama 4 language model earlier this year, the company revamped its approach to lead the competition.

Media reports indicate that Meta has hired numerous senior executives from Apple, Anthropic, and OpenAI into its Superintelligence Labs. Zuckerberg reportedly offered $100 million‑level packages to attract top minds.