Meta Prometheus supercluster advances toward million-GPU operational scale

Assume that by March 31, 2027, Meta materially advances the Prometheus supercluster in New Albany, Ohio toward roughly 1 million Nvidia GPUs through concrete implementation evidence rather than only target language. The initiating change is that Meta, Nvidia, power or data-center counterparties, permitting/interconnection records, or reliable business and technology reporting confirms that Nvidia accelerator systems have been installed, made operational, delivered under Meta control for installation, or supported by energized data-center and power infrastructure sufficient to put Prometheus at or near the planned million-GPU scale. The reported system mix may include GB200, GB300, Vera Rubin, or other Nvidia systems; exact GPU counts, power draw, rack counts, commissioning dates, and utilization levels remain unavailable unless later disclosed.