CoreWeave LiquidLab is a large GPU cluster operated by CoreWeave, located in the United States, in the private sector.
It is built on 4,000 NVIDIA H100 SXM5 80GB. That amounts to about 4K H100e of AI compute. The cluster draws roughly 5.61 MW. Its hardware is estimated at $188M.
Epoch AI rates the certainty of this record as likely. Unclear exactly what type of GPU they use. Likely operational by now