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

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

Rig A

AWS · cloud-model

Change

Rig B

NVIDIA · workstation

Change

0 of 42 fields comparable

Spec	AWS Trainium2	NVIDIA RTX PRO 6000 Blackwell
AI Tops	—	4,000
Bandwidth GB S	—	1,792
Bf16 Tflops PER Chip	667	—
Chips PER Trn2 Instance	16	—
Chips PER Ultraserver	64	—
Cuda Cores	—	24,064
Displayport Count	—	4
Displayport Version	—	2.1
ECC	—	true
Form factor	—	5.4 in H x 12.0 in L, dual slot
FP8 Dense Petaflops PER Chip	1.3	—
FP8 Sparse Petaflops PER Chip	5.2	—
HBM Bandwidth TB S PER Chip	2.9	—
HBM Capacity GB PER Chip	96	—
HBM Type	HBM3	—
Memory BUS Bits	—	512
Memory Speed Gbps	—	28
Neuroncores PER Chip	8	—
Nvdec Count	—	4
Nvdec GEN	—	6
Nvenc Count	—	4
Nvenc GEN	—	9
Pcie GEN	—	5
RT Cores	—	188
RT Cores GEN	—	4
Slot Width	—	2
Streaming Multiprocessors	—	188
TDP (W)↓ better	—	600
Tensor Cores	—	752
Tensor Cores GEN	—	5
Trainium3 GA	2025-12-02	—
Trainium3 HBM Bandwidth TB S	4.9	—
Trainium3 Hbm3e GB PER Chip	144	—
Trn2 Instance FP8 Petaflops	20.8	—
Trn2 Instance HBM TB	1.5	—
Trn3 Ultraserver FP8 Pflops	362	—
Trn3 Ultraserver MAX Chips	144	—
Ultraserver FP8 Petaflops	83.2	—
Ultraserver Total HBM Bandwidth TB S	185	—
Ultraserver Total HBM TB	6	—
VRAM (GB)	—	96
Vram Type	—	GDDR7 ECC