NVIDIA Blackwell B200
The most powerful AI infrastructure for autonomous driving
72 PetaFLOPS of Processing Power
Our project rises on an 8x NVIDIA B200 GPU cluster, marking a turning point in the autonomous driving world. With this infrastructure, we don't just process data – we simulate the physical world in real-time.
Why Blackwell B200
Unprecedented Speed
The B200's massive parallel processing capability allows us to run complex world simulations at speeds previously thought impossible.
FP4 Precision
New FP4 precision enables efficient training of massive models while maintaining the accuracy required for safety-critical applications.
NVLink Interconnect
High-bandwidth NVLink connectivity enables our 8 GPUs to work as a single, unified computing entity for maximum efficiency.
Explore Our Software Architecture
See how we leverage this hardware power with our Generative World Models.
World Models