Planned Unlikely

Project Ceiba Phase 1

Amazon,NVIDIA • United States of America
Total GPUs
16.4K
H100 Equiv
16.4K
Power
23.0 MW
FP16
16.2K PF

System Equivalents

This datacenter's 16.4K GPUs equal:

NVIDIA HGX H200 (8-GPU)
8 GPUs/system
2.0K
NVIDIA HGX H200 (4-GPU)
4 GPUs/system
4.1K

Specifications

Primary GPUNVIDIA GH200
GPU Count16.4K
SectorPrivate

Additional Information

The system will feature 16,384 GH200 NVL32 Grace Hopper Superchips and is expected to be the world’s largest cloud AI supercomputer.

This plan has been updated to use Blackwell chips. Unclear if this supercomputer was ever completed