DEPLOYDatabase

Chip comparison

NVIDIA DRIVE Thor vs NVIDIA GB200 Grace Blackwell Superchip

Side-by-side specs

Straight from each vendor's datasheet. Winner in bold: higher throughput, memory, or transistor count; lower power draw or price. Sparse (2:4) throughput roughly doubles the FP8/INT8 numbers shown.

FieldNVIDIA DRIVE ThorNVIDIA GB200 Grace Blackwell Superchip
Process nodeTSMC 4NPTSMC 4NP
TDP130 W2700 W
Memory128 GB LPDDR5X384 GB HBM3e
Memory bandwidth16,000 GB/s
FP8 (dense)2,000 TFLOPS9,000 TFLOPS
INT8 (dense)2,000 TOPS
Form factorAuto-boardSuperchip (2xB200 + 1xGrace)
Announced2022-09-202024-03-18
Released2025-06-012024-12-01

Robots running NVIDIA GB200 Grace Blackwell Superchip

No robots publicly running it yet.

Data centers with NVIDIA DRIVE Thor

No data centers publicly running it yet.

Common questions

NVIDIA DRIVE Thor vs NVIDIA GB200 Grace Blackwell Superchip: which is faster for training?

NVIDIA GB200 Grace Blackwell Superchip has 4.50x the dense FP8 throughput of the other (NVIDIA DRIVE Thor: 2,000 TFLOPS; NVIDIA GB200 Grace Blackwell Superchip: 9,000 TFLOPS). Real-world training speed also depends on memory bandwidth, interconnect, and software stack: see the spec diff for those inputs.

NVIDIA DRIVE Thor vs NVIDIA GB200 Grace Blackwell Superchip: which has more memory?

NVIDIA GB200 Grace Blackwell Superchip carries more HBM (NVIDIA DRIVE Thor: 128 GB LPDDR5X; NVIDIA GB200 Grace Blackwell Superchip: 384 GB HBM3e). Memory capacity sets the largest model that fits on one chip without partitioning; memory bandwidth sets inference throughput.

NVIDIA DRIVE Thor vs NVIDIA GB200 Grace Blackwell Superchip: which is more power-efficient?

Dense FP8 TFLOPS per watt: NVIDIA DRIVE Thor 15.38 (2,000 ÷ 130 W); NVIDIA GB200 Grace Blackwell Superchip 3.33 (9,000 ÷ 2700 W). NVIDIA DRIVE Thor wins at the die level; system-level efficiency also depends on cooling and interconnect.

NVIDIA DRIVE Thor vs NVIDIA GB200 Grace Blackwell Superchip: which is more widely used?

NVIDIA DRIVE Thor: 11 robots and 0 data centers publicly running it. NVIDIA GB200 Grace Blackwell Superchip: 0 robots and 11 data centers. Empty here means no operator has said publicly, not zero adoption in the market.

See also: every chip comparison · NVIDIA DRIVE Thor full page · NVIDIA GB200 Grace Blackwell Superchip full page.