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.

0 of 39 fields comparable

⇄ swap sides
SpecNVIDIA H200Tesla Optimus
AI Stack—Tesla FSD-derived end-to-end neural networks (vision-only)
Battery KWH—2.3
BF16 TFLOPS (sparse)1,979—
Form factorSXM—
FP16 TFLOPS (sparse)1,979—
FP32 TFLOPS67—
FP64 TFLOPS34—
FP8 TFLOPS (sparse)3,958—
Fremont Line Capacity Robots PER Year—1,000,000
Fremont Status Q2 2026—S/X lines decommissioned; Optimus lines being installed
Gen2 Announcement Date—2023-12-13
Gen2 Hand DOF—11
Gen2 Height CM—173
Gen2 Neck DOF—2
Gen2 Total DOF Unverified—28
Gen2 Walking Speed Boost PCT—30
Gen2 Weight KG—57
Gen2 Weight Reduction KG—10
Gen3 Announcement Quarter—Q1 2026
Gen3 Forearm Actuators—25
Gen3 Hand DOF—22
Gen3 Payload KG—20
Gen3 Production Target Superseded—summer 2026
Gen3 Total Actuators PER Robot—50
Gen3 Unveil Status—not unveiled as of 2026-09-23 (Tesla North)
HBM (GB)141—
INT8 TOPS (sparse)3,958—
Memory bandwidth (TB/s)4.8—
Memory typeHBM3e—
MIG Instances7—
MIG PER Instance GB18—
NVLink bandwidth (GB/s)900—
Payload PER Hand KG—9
Pcie Gen5 Bandwidth GB S128—
Public Sale Target—end of 2027 (Musk at Davos, 2026-01-22)
Runtime Hours Light Duty—8
Status Checked—2026-09-24
TDP max (W)↓ better700—
Walking Speed KM H—8