Amazon Titan training cluster

Compute

4.3K H100e

Chips

13,760

First operational

Jan 15, 2023

Amazon Titan training cluster is a large GPU cluster operated by Amazon, located in the United States, in the private sector. It came online around January 2023. At that moment it ranked number 6 among known clusters.

It is built on 13,760 NVIDIA A100. That amounts to about 4.3K H100e of AI compute. The cluster draws roughly 11.42 MW. Its hardware is estimated at $310M.

Epoch AI rates the certainty of this record as likely. Model was released to large clients April 2023, took 48 days to train, so this cluster was operational at least three months beforehand

Full record

Owner: Amazon
Users: Amazon
Sector: Private
Status: Existing
Country: United States
Primary chip: NVIDIA A100
Primary chip count: 13,760
Total AI chips: 13,760
H100 equivalents: 4.3K
8-bit throughput: 8.6×10¹⁸ OP/s
Power capacity: 11.42 MW
Hardware cost: $310M
First operational: Jan 15, 2023
Rank when it came online: 6
Certainty: Likely