B300
288GB HBM3e per GPU
View GPU All GPUsB200
Blackwell architecture180GB HBM3e per GPU
GPU Catalog / NVIDIA
B200
GPU infrastructure.
A reserve-only Blackwell GPU for large-model training, inference, and accelerated AI pipelines.
- Architecture
- Blackwell
- GPU memory
- 180GB HBM3e per GPU
- Availability
- Reservation only
- Reservation rate
- Pricing on request
- Lead time
- 3-7 weeks
Reservation
Reserve capacity.
Estimated lead time is 3-7 weeks, subject to availability and shipping.
- Access
- Reservation only
- Rate
- Pricing on request
- Estimated lead time
- 3-7 weeks
- Qualification
- Subject to availability and shipping.
Hardware profile
Where B200 fits.
NVIDIA B200 uses the Blackwell architecture with 180GB of HBM3e memory per GPU. NVIDIA positions B200 systems for develop-to-deploy AI pipelines, including large language models, recommender systems, and chatbots.
Choose B200 when the workload needs Blackwell-generation AI compute and a 180GB memory envelope per GPU, with deployment planned around a reservation lead time.
NVIDIA product information- 01
Large-model inference
Reserve Blackwell infrastructure for language-model serving and inference workloads with large memory requirements.
- 02
AI training
Build planned training capacity for transformer models and other compute-intensive neural networks.
- 03
Develop-to-deploy pipelines
Use one GPU generation across model development, fine-tuning, evaluation, and production inference.
Deployment sizing
Size before you provision.
Hardware fit depends on the model, precision, batch size, software stack, and total working set. Published use cases are not performance guarantees.
- 01 / Memory
Confirm the working set
Include weights, caches, activations, batches, and framework overhead.
- 02 / Software
Validate the stack
Check CUDA, driver, framework, precision, and container requirements.
- 03 / Capacity
Plan the lead time
Allow 3-7 weeks for availability and shipping before the planned deployment.