Your NVIDIA Blackwell Journey Starts Here
In this transformative moment of AI, where the evolving scaling laws continue to push limits of data center capabilities, our latest NVIDIA Blackwell-powered solutions, developed through close collaboration with NVIDIA, offers unprecedented computational performance, density, and efficiency with next-generation air-cooled and liquid-cooled architecture. With our readily deployable AI Data Center Building Block solutions, Supermicro is your premier partner to start your NVIDIA Blackwell journey, providing sustainable, cutting-edge solutions that accelerate AI innovations.
End-to-End AI Data Center Building Block Solutions Advantage
A broad range of air-cooled and liquid-cooled systems with multiple CPU options, a full data center management software suite, turn-key rack level integration with full networking, cabling, and cluster level L12 validation, global delivery, support and service.
- Vast Experience
- Supermicro’s AI Data Center Building Block Solutions power the largest liquid-cooled AI Data Center deployment in the World.
- Flexible Offerings
- Air or liquid-cooled, GPU-optimized, multiple system and rack form factors, CPUs, storage, networking options. Optimizable for your needs.
- Liquid-Cooling Pioneer
- Proven, scalable, and plug-and-play liquid-cooling solutions to sustain the AI revolution. Designed specifically for NVIDIA Blackwell Architecture.
- Fast Time to Online
- Accelerated delivery with global capacity, world class deployment expertise, one-site services, to bring your AI to production, fast.
Next-Gen Liquid-Cooled System
Up to 96 NVIDIA HGX™ B200 GPUs in a Single Rack with Maximum Scalability and Efficiency
The new liquid-cooled 4U NVIDIA HGX B200 8-GPU system features newly developed cold plates and advanced tubing design paired with the new 250kW coolant distribution unit (CDU) more than doubling the cooling capacity of the previous generation in the same 4U form factor. The new architecture further enhances efficiency and serviceability of the predecessor that are designed for NVIDIA HGX H100/H200 8-GPU. Available in 42U, 48U or 52U configuration, the rack scale design with the new vertical coolant distribution manifolds (CDM) means that horizontal manifolds no longer occupy valuable rack units. This enables 8 systems, 64 NVIDIA Blackwell GPUs in a 42U rack and all the way up to 12 systems with 96 NVIDIA GPUs in a 52U rack.
4U Liquid-cooled System
for NVIDIA HGX B200 8-GPU
Maximum Density, Performance and Efficiency with Next-Gen Liquid-cooling for frontier model training and Inference
Air-Cooled System, Evolved
The Best Selling Air-cooled System Redesigned and Optimized for NVIDIA HGX B200 8-GPU
The new air-cooled 10U NVIDIA HGX B200 system features a redesigned chassis with expanded thermal headroom to accommodate eight 1000W TDP Blackwell GPUs. Up to 4 of the new 10U air-cooled systems can be installed and fully integrated in a rack, the same density as the previous generation, while providing up to 15x inference and 3x training performance. All Supermicro NVIDIA HGX B200 systems are equipped with a 1:1 GPU-to-NIC ratio supporting NVIDIA BlueField®-3 or NVIDIA ConnectX®-7 for scaling across a high-performance compute fabric.
10U Air-cooled System
for NVIDIA HGX B200 8-GPU
Industry-leading Air-cooling Design for large language model training and high volume inference
An Exascale of Compute in a Rack
End-to-end Liquid-cooling Solution for NVIDIA GB200 NVL72
Supermicro’s GB200 NVL72 solution represents a breakthrough in AI computing infrastructure, combining Supermicro’s end-to-end liquid-cooling technology. It enables up to 25x performance improvement at the same power envelope compared to previous NVIDIA Hopper generations, while reducing data center electricity costs by up to 40%. The system integrates 72 NVIDIA Blackwell GPUs and 36 NVIDIA Grace CPUs in a single rack, delivering exascale computing capabilities through NVIDIA’s most extensive NVLink™ network to date, achieving 130 TB/s of GPU communications.
The 48U solution’s versatility supports both liquid-to-air and liquid-to-liquid cooling configurations, accommodating various data center environments.
NVIDIA GB200 NVL72 SuperCluster
for NVIDIA GB200 Grace™ Blackwell Superchip
72 NVIDIA Blackwell GPUs in one NVIDIA NVLink domain. The apex of AI computing architecture.
Liquid-to-Air Solution for NVIDIA GB200 NVL72
Plug-and-Play Scalable Unit Readily Deployable for NVIDIA Blackwell
Available in 42U, 48U, or 52U rack configurations for Air-cooled or liquid-cooled data centers, the new SuperCluster designs incorporate NVIDIA Quantum InfiniBand or NVIDIA Spectrum™ networking in a centralized rack, the liquid-cooled SuperClusters enabling a non-blocking, 256-GPU scalable unit in five 42U/48U racks or an extended 768-GPU scalable unit in nine 52U racks for the most advanced AI data center deployments. Supermicro also offers an in-row CDU option for large deployments, as well as liquid-to-air cooling rack solution that doesn't require facility water. The air-cooled SuperCluster design follows the proven, industry leading architecture of the previous generation, providing a 256-GPU scalable unit in nine 48U racks.
End-to-end Data Center Building Block Solution and Deployment Services for NVIDIA Blackwell
Supermicro serves as a comprehensive one-stop solution provider with global manufacturing scale, delivering data center-level solution design, liquid-cooling technologies, switching, cabling, a full data center management software suite, L11 and L12 solution validation, onsite installation, and professional support and service. With production facilities across San Jose, Europe, and Asia, Supermicro offers unmatched manufacturing capacity for liquid-cooled or air-cooled rack systems, ensuring timely delivery, reduced total cost of ownership (TCO), and consistent quality.
Supermicro’s NVIDIA Blackwell solutions are optimized with NVIDIA Quantum InfiniBand or NVIDIA Spectrum™ networking in a centralized rack for the best infrastructure scaling and GPU clustering, enabling a non-blocking, 256-GPU scalable unit in five racks or an extended 768-GPU scalable unit in nine racks. This architecture, with native support for NVIDIA Enterprise software, combines with Supermicro’s expertise in deploying the world’s largest liquid-cooled data centers to deliver superior efficiency and unmatched time-to-online for today's most ambitious AI data center projects.
NVIDIA Quantum InfiniBand and Spectrum Ethernet
From mainstream enterprise servers to high-performance supercomputers, the NVIDIA Quantum InfiniBand and Spectrum™ networking technology activate the most scalable, fastest, and secure end-to-end networking.
NVIDIA SuperNICs
Supermicro is at the forefront of adopting NVIDIA SuperNICs: NVIDIA ConnectX for InfiniBand and NVIDIA BlueField-3 SuperNIC for Ethernet. All of Supermicro’s NVIDIA HGX B200 systems are equipped with 1:1 networking to each GPU to enable NVIDIA GPUDirect RDMA (InfiniBand) or RoCE (Ethernet) for massively parallel AI computing.
NVIDIA AI Enterprise Software
Full access to NVIDIA application frameworks, APIs, SDKs, toolkits and optimizers, along with ability to deploy AI blueprints, NVIDIA NIMs, RAGs, and the latest optimized AI foundation models. NVIDIA AI Enterprise Software streamlines development and deployment of production-grade AI applications with enterprise-grade security, support and stability to ensure smooth transition from prototype to production.