Meta announces several multi-gigawatt data centers, first planning to come online in 2026

Meta's AI Infrastructure Expansion - Meta is constructing several multi-gigawatt compute clusters, with the first, Prometheus, expected to be online in 2026 and capable of 1 gigawatt of energy [1] - Meta is also building Hyperion, which will scale up to 5 gigawatts over several years, alongside multiple Titan clusters [2] - This energy capacity rivals that of hyperscalers like Microsoft and Amazon and matches the scale OpenAI and Oracle are aiming for with Project Stargate [2] Strategic Implications - Meta's investment signals its commitment to matching the ambition of other AI leaders "watt for watt" [3] - Zuckerberg's significant spending on AI aims to attract AI talent and dominate the next era of frontier AI models [3] - Meta's massive compute power aims to capture attention from hyperscalers, despite not having an enterprise cloud business, to serve its own AI ambitions [5] Market Reaction - Meta's stock price initially increased by approximately 0.9% following the announcement [3]