Skip to content
MCS · Rigs · Compare

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

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

Rig A

Groq Language Processing Unit

Groq · accelerator

Change

Rig B

AWS Trainium2

AWS · cloud-model

Change

0 of 33 fields comparable

⇄ swap sides
SpecGroq Language Processing UnitAWS Trainium2
Bf16 Tflops PER Chip—667
Chip TO Chip Links11—
Chips PER Trn2 Instance—16
Chips PER Ultraserver—64
DOJ Probe Reported2026-09-10—
Form factorPCIe Gen4 x16—
FP8 Dense Petaflops PER Chip—1.3
FP8 Sparse Petaflops PER Chip—5.2
Groq3 LPX Full Production2026-08-24—
Groqcard Sales Ended Reported2024-04-05—
HBM Bandwidth TB S PER Chip—2.9
HBM Capacity GB PER Chip—96
HBM Type—HBM3
Neuroncores PER Chip—8
Nvidia Consideration AT Closing USD Billion13—
Nvidia Consideration Deferred USD Billion4—
Nvidia License Agreement2025-12-24—
ON Chip Sram MB230—
ON DIE Memory Bandwidth TB S80—
Peak Fp16 Tflops188—
Peak Int8 Tops750—
Process Node14nm—
Status Checked2026-09-24—
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