Inference Stack
The reference for how AI infrastructure actually fits together
Architectures
Catalog
Where it runs
About
Back to the interactive map
Silicon
B200
nvidia
Blackwell-architecture GPU — 192 GB HBM3e, 8 TB/s memory bandwidth
Official docs
View in catalog
Used by
CUDA
Leads to
AWS p6 (B200)
GCP a3-ultra (B200)