Meta is accelerating its AI infrastructure strategy with the development of Hyperion, a new data center project expected to deliver five gigawatts of computational capacity to its AI research lab. CEO Mark Zuckerberg announced the plan via Threads, describing Hyperion’s footprint as large enough to rival the size of Manhattan. The facility will be located in Richland Parish, Louisiana, where Meta previously committed $10 billion to data infrastructure.
This move follows Meta’s recruitment of top AI talent, including Alexandr Wang and Daniel Gross, to lead its Superintelligence Lab. A separate 1GW supercluster named Prometheus will come online in 2026 in New Albany, Ohio, positioning Meta among the first tech giants to control AI infrastructure at this scale.
Together, Hyperion and Prometheus aim to elevate Meta’s competitiveness against OpenAI, Google DeepMind, and Anthropic. However, the scale of energy demand — projected to rival the consumption of millions of homes — raises growing concerns about the environmental impact and strain on local utilities.
While Meta pushes ahead, it joins a wave of hyperscalers — including xAI’s Colossus, OpenAI’s Stargate, and CoreWeave — building energy-intensive infrastructure for frontier AI. Backed by growing political support, including from President Donald Trump and U.S. Energy Secretary Chris Wright, the race to power the next era of intelligence may reshape America’s energy economy. Data centers could consume up to 20% of U.S. energy by 2030, dramatically up from 2.5% in 2022.