Meta CEO Zuckerberg says first AI data supercluster will come online in 2026

On Monday, Meta's CEO Mark Zuckerberg unveiled plans to invest hundreds of billions of dollars into the development of artificial intelligence computing infrastructure. He confirmed that the company aims to have its inaugural supercluster operational by next year. This supercluster will serve as a sophisticated computing network specifically designed for training advanced AI models and managing their extensive workloads. Zuckerberg shared on Facebook that the newly formed Meta Superintelligence Labs will boast industry-leading computational capabilities, offering the highest compute resources per researcher. "I'm eager to collaborate with top researchers to push the boundaries of AI technology!" he expressed. The first of these superclusters is named Prometheus, with plans for several additional multi-gigawatt clusters. One notable cluster, dubbed Hyperion, is projected to expand its capacity to five gigawatts over the coming years. Recently, Zuckerberg has been aggressively hiring AI talent, exemplified by a significant $14 billion investment in Scale AI. In June, he announced the creation of Meta Superintelligence Labs, a team composed of elite AI researchers and engineers. This initiative comes in response to Zuckerberg's concerns regarding Meta's AI advancements, particularly after the release of the Llama 4 AI models in April did not meet developer expectations. He is now revamping the company’s AI strategy to better position Meta against competitors like OpenAI and Google. "My priority for our superintelligence initiative is to assemble the most talented and skilled team in the industry," Zuckerberg concluded.