Meta Plans AI Supercluster Deployment Across Multiple US Sites by 2026

Meta CEO Mark Zuckerberg announced a major investment in artificial intelligence infrastructure, revealing plans to launch the company's first AI data supercluster, Prometheus, in 2026. This initiative is part of a broader strategy involving hundreds of billions of dollars in capital expenditures focused on building massive data centers and recruiting top AI talent to compete with industry leaders like OpenAI, Google, and Microsoft. Meta is constructing multiple multi-gigawatt AI clusters, including Hyperion, which may scale to five gigawatts, exceeding the power consumption of some small countries. In an effort to accelerate deployment, Meta has even utilized temporary data center setups in tents, despite the risks, highlighting the urgency behind their AI ambitions. Zuckerberg emphasized the creation of Meta Superintelligence Labs, aiming for industry-leading compute capacity per researcher and assembling an elite, talent-dense team with lucrative recruitment efforts. These investments reflect Meta's response to internal frustrations with AI progress and its commitment to advancing toward superintelligence, a level of AI surpassing human capabilities.