AI Chips & Infrastructure
Meta Details Prometheus AI Supercluster in Ohio
The first phase has trained Muse models; Meta says the cluster is scaling toward roughly one million GPUs.
Illustration of GPU server racks and cooling infrastructure inside a large data center.
Photo: Kronos News
Meta says the first phase of Prometheus, its gigawatt-scale AI supercluster in New Albany, Ohio, is complete and was used to train its Muse foundational models. The company says the facility now spans hundreds of thousands of GPUs and is being scaled toward roughly one million. [1][2]
Meta describes Prometheus as a network of interconnected buildings operating as one system. It currently uses NVIDIA GB200 and GB300 GPU platforms, with a Vera Rubin-based platform planned; Meta says the larger cluster is intended to support future model training. [1][2]
The performance figures and expansion target are company-reported, not an independent assessment. Meta’s post describes infrastructure and training operations, but does not establish that the cluster or models will deliver particular real-world outcomes. [1]
Editorial notes
Transparency note
AI assisted drafting. Human edited and reviewed.
- AI assisted
- Yes
- Human review
- Yes
- Last updated
Risk assessment
The main infrastructure details and expansion target come from Meta’s own account; the secondary source reports those claims but does not independently verify them.
Sources
Related stories
View all
Topics
About the author
Kronos News Desk covers ai chips & infrastructure and editorial analysis for Kronos News.