Aramco Groq Inference Cluster is a large GPU cluster operated by Saudi Aramco, located in Saudi Arabia, in the private sector. It came online around January 2025. At that moment it ranked number 28 among known clusters.
It is built on 19,725 GroqChip LPU v1. That amounts to about 7.5K H100e of AI compute. The cluster draws roughly 8.49 MW. Its hardware is estimated at $675M.
Epoch AI rates the certainty of this record as likely. This cluster is designed for only inference, not training