Side-by-side, with the numbers that matter.

Two rigs, one table. Higher numbers win in cobalt; TDP and price flip.

Rig A

NVIDIA · workstation

Change

Rig B

AMD · accelerator

Change

0 of 42 fields comparable

Spec	NVIDIA Eos DGX SuperPOD	AMD Instinct MI355X
AI Exaflops FP8	18.4	—
Bf16 Matrix Pflops Dense	—	2.5
Bf16 Matrix Pflops Sparse	—	5
Compute Units	—	256
Cooling	—	direct liquid cooling (air-cooled option available)
DGX Node Count	576	—
DGX PER POD	32	—
Form factor	—	OAM
Fp16 Matrix Pflops Dense	—	2.5
Fp16 Matrix Pflops Sparse	—	5
Fp32 Vector Tflops	—	157.3
Fp64 Matrix Tflops	—	78.6
FP8 Matrix Pflops Dense	—	5
FP8 Matrix Pflops Sparse	—	10.1
GPU Count	4,608	—
GPU Type	NVIDIA H100 (Hopper)	—
Gpus PER Node	8	—
HBM (GB)	—	288
HBM Type	—	HBM3E
Helios Gpus PER Rack	—	72
HPC Petaflops Fp64	275	—
Interconnect	NVIDIA Quantum-2 InfiniBand	—
Location	NVIDIA (in-house)	—
Memory bandwidth (TB/s)	—	8
Mxfp4 Matrix Pflops	—	10.1
Mxfp6 Matrix Pflops	—	10.1
Nvlink Switches	360	—
Pcie GEN	—	5
Pods	18	—
Process Node NM	—	3
Quantum2 IB Switches	500	—
Rmax Pflops	121.4	—
Rpeak Pflops	188.65	—
Successor Cdna Version	—	5
Successor Family	—	AMD Instinct MI400 Series (MI455X)
Successor Launch	—	2026-07-23
Successor Mi455x Bandwidth TB S	—	23.3
Successor Mi455x Hbm4 GB	—	432
TBP Watts	—	1,400
Top500 Rank Jun2024	10	—
Top500 Rank Jun2026	23	—
Top500 Rank Nov2023	9	—