Meta's Prometheus AI supercluster in Ohio completes first phase, aims for 1 million GPUs
Overview
Meta says the first phase of its gigawatt-scale Prometheus supercluster in New Albany, Ohio, is complete and was used to train its Muse models.
The company says it plans roughly 1 million distributed GPUs working in lockstep at full scale.
In an Engineering at Meta post, the company says interruptions fell 99% from their mid-2025 peak through firmware fixes, telemetry, and work with NVIDIA, and that mean time between interruptions rose to more than 54,000 GPU hours.
Written by AI from the articles below · updated Oct 19, 9:23 AM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- Engineering at MetaOfficialPickMeta details how it built and stabilized its Prometheus AI supercluster
AIMeta says the first phase of its gigawatt-scale Prometheus supercluster in New Albany, Ohio, is complete and was used to train its Muse models, with a goal of roughly 1 million GPUs. The company reports that interruptions fell 99% from their mid-2025 peak through firmware fixes, telemetry, and cross-partner work with NVIDIA, and that MTBI rose to more than 54K GPU hours.
Heat Trend
Not enough continuous observations to show a trend yet.