NVIDIADGX B200
Eight NVIDIA Blackwell GPUs interconnected with fifth-generation NVIDIA® NVLink®, DGX B200 delivers leading-edge performance, offering 3X the training performance and 15X the inference performance of previous generations.
Available at the most cost-effective pricing
Launch your AI products faster with on-demand GPUs and a global network of data center partners
Bare metal
Complete control over a physical machine providing absolute performance & control
from $0.00/hr
- No noisy neighbours
- SpectrumX local networking
- 300Gbps external connectivity
- NVMe SSD storage
Looking to scale? Please contact us for enterprise solutions.
Speak with an expertThe NVIDIA DGX B200 is perfect for a wide range of workloads
Deploying AI based workloads on CUDO Compute is easy and cost-effective. Follow our AI related tutorials.
Specifications
| Starting from | Contact us for pricing |
| Architecture | NVIDIA Blackwell |
| GPU | 8x NVIDIA Blackwell GPUs |
| GPU Memory | 1,440GB total GPU memory |
| Performance | 72 petaFLOPS training and 144 petaFLOPS inference |
| CPU | 2 Intel® Xeon® Platinum 8570 Processors - 112 Cores total, 2.1 GHz(Base), 4 GHz(Max Boost) |
| System Memory | Up to 4TB |
| Networking | 4x OSFP ports serving 8x single-port NVIDIA ConnectX-7 VPI | 2x dual-port QSFP112 NVIDIA BlueField-3 DPU |
| Management Network | 10Gb/s onboard NIC with RJ45, 100Gb/s dual-port ethernet NIC, Host baseboard management controller (BMC) with RJ45 |
| Storage | OS: 2x 1.9TB NVMe M.2, Internal storage: 8x 3.84TB NVMe U.2 |
| Software | NVIDIA AI Enterprise: Optimized AI Software, NVIDIA Base Command™: Orchestration, Scheduling, and Cluster Management, DGX OS / Ubuntu: Operating system |
Use cases
Powerhouse of AI Performance
Powered by the NVIDIA Blackwell architecture’s advancements in computing, DGX B200 delivers 3X the training performance and 15X the inference performance of DGX H100.
Real Time Large Language Model Inference
Token-to-token latency (TTL) = 50ms real time, first token latency (FTL) = 5s, input sequence length = 32,768, output sequence length = 1,028, 8x eight-way DGX H100 GPUs air-cooled vs. 1x eight-way DGX B200 air-cooled, per GPU performance comparison.
High-performance computing
From complex scientific simulations to weather forecasting and intricate financial modeling, the DGX B200 will empower diverse organizations to accelerate high-performance computing tasks. Its unmatched memory bandwidth and processing capabilities ensure smooth operation for workloads of any scale, allowing you to achieve unmatched results faster than ever.
Browse alternative GPU solutions for your workloads
Access a wide range of performant NVIDIA and AMD GPUs to accelerate your AI, ML & HPC workloads
NVIDIA GB200 NVL72
Get the highest performing GB200 NVL72 GPUs at scale on our reserved cloud.
NVIDIA H200
Get the highest performing H200 GPUs at scale on our reserved cloud.
NVIDIA B100
Get the highest performing B100 GPUs at scale on our reserved cloud.
NVIDIA RTX A4000 SFF Ada
Quickly deploy RTX A4000 SFF Ada GPUs on our on-demand cloud.
AMD MI250/300
Get the highest performing MI250/300 GPUs at scale on our reserved cloud.
An NVIDIA preferred partner for compute
We're proud to be an NVIDIA preferred partner for compute, offering the latest GPUs and high-performance computing solutions.
Also trusted by our other key partners:
Pricing & reservation enquiry
Enquire about access today to test the DGX B200 GPU Cloud, or reserve your DGX B200 Cloud on CUDO Compute for as long as you want it, with unique contracts tailored to suit your needs.
Get started today or speak with an expert...
Available Mon-Fri 9am-5pm UK time