Nvidia GTC DC 2025: Supermicro Expands Collaboration with Nvidia

Summary:

Deepening collaboration with Nvidia to bring Nvidia Vera Rubin NVL144 and Nvidia Vera Rubin CPX platforms to market in 2026 to provide more than 3x AI attention acceleration vs. Blackwell Ultra
Powered by Data Center Building Block Solutions, Supermicro introduces a compact 2U Nvidia HGX B300 8-GPU system with OCP-based rack-scale design supporting up to 144 GPUs in a single rack
Supermicro’s San Jose, CA manufacturing ensures solutions are developed, constructed, and validated in the USA, are TAA-compliant, and are Buy American Act-capable
Supporting a full-stack Nvidia AI Factory for Government reference design with expansion of AI solution portfolio, including deskside AI supercomputer, Super AI Station based on Nvidia GB300 and the new rack-scale Nvidia GB200 NVL4 HPC solutions

At Nvidia GTC 2025 in Washington, D.C, Super Micro Computer, Inc. is showcasing its advanced AI infrastructure solutions, highlighting systems tailored to meet the stringent requirements of federal customers.

The company announced its plans to deliver next-gen Nvidia AI platforms, including the Nvidia Vera Rubin NVL144 and Vera Rubin NVL144 CPX in 2026. Additionally, the firm introduces U.S.-manufactured, TAA (Trade Agreements Act)-compliant systems, including the high-density 2U Nvidia HGX B300 8-GPU system with up to 144 GPUs/rack and an expanded portfolio featuring a Super AI Station based on Nvidia GB300 and the new rack-scale GB200 NVL4 HPC solutions.

“Our expanded collaboration with Nvidia and our focus on U.S.-based manufacturing position Supermicro as a trusted partner for federal AI deployments. With our corporate headquarters, manufacturing, and R&D all based in San Jose, California, in the heart of Silicon Valley, we have an unparalleled ability and capacity to deliver first-to-market solutions are developed, constructed, validated (and manufactured) for American federal customers,” said Charles Liang, president and CEO, Supermicro. “The result of many years of working hand-in-hand with our close partner Nvidia – also based in Silicon Valley – Supermicro has cemented its position as a pioneer of American AI infrastructure development.“

Supermicro is expanding its latest solutions based on Nvidia HGX B300 and B200, GB300 and GB200, and RTX PRO 6000 Blackwell Server Edition GPUs that provide compute performance, efficiency, and scalability for key federal government workloads such as cybersecurity and risk detection, engineering and design, healthcare and life sciences, data analytics and fusion platforms, modeling and simulation, and secure virtualized infrastructure.

The company’s focus on U.S.-based manufacturing is a cornerstone of its business focus. All government-optimized systems are developed, constructed, and rigorously validated at its global headquarters in San Jose, CA, ensuring full compliance with the TAA and eligibility under the Buy American Act. This domestic production capability enhances supply chain security and meets federal requirements for trusted, high-quality technology solutions.

Through its collaboration with Nvidia, Supermicro is set to introduce Nvidia Vera Rubin NVL144 and Rubin CPX platforms in 2026. These platforms will deliver exceptional AI training and inference performance of their predecessors, empowering organizations to handle complex AI workloads with efficiency.

The company is also unveiling its most compact system, a 2U Nvidia HGX B300 8-GPU server, featuring an OCP-based rack-scale design powered by Supermicro Data Center Building Block Solutions. This architecture supports up to 144GPUs in a single rack, delivering outstanding performance and scalability for large-scale AI and HPC deployments in government data centers.

The firm is expanding its government-focused portfolio by optimizing for Nvidia AI Factory for Government reference design. The Nvidia AI Factory for Government is a full-stack, end-to-end reference design that provides guidance for deploying and managing multiple AI workloads on-premises and in the hybrid cloud while meeting the compliance needs of high-assurance organizations.

The portfolio now includes the Super AI Station based on Nvidia GB300, and the rack-scale Nvidia GB200 NVL4 HPC solutions, both optimized for federal environments with enhanced security, reliability, and scalability to meet stringent government standards.

Support for New Accelerated Networking
Further highlighting Supermicro’s first-to-market development with new Nvidia technologies, it also announced support for the newly-announced BlueField-4 DPU and ConnectX-9 SuperNIC in gigascale AI factories. When available, these new accelerated infrastructure technologies will be readily integrated into new the company’s AI systems to provide faster cluster-scale AI networking, storage access and data processing offload for the next-gen of Nvidia AI infrastructure. Supermicro’s modular hardware design will enable the rapid integration of new technologies such as the Nvidia BlueField-4 and ConnectX-9 into existing systems designs with minimal re-engineering, speeding up time-to-market and reducing development costs.

Super AI Station brings AI server power to desktop
Continuing its record of first-to-market implementation of new Nvidia technologies, the company announces the new liquid-cooled ARS-511GD-NB-LCC Super AI Station. By bringing the high-end server grade GB300 Superchip into a deskside form factor, this platform is the first of its kind, unleashing unparalleled performance, and resulting in more than 5x AI PFLOPS of computing power, compared to traditional PCIe based GPU workstations. This new Super AI Station is a complete solution for AI model training, fine-tuning, applications and algorithms prototyping and development, that can be deployed on-prem for unmatched latency and full data security, supporting models up to 1 trillion parameters. This self-contained platform is for government agencies, startups, deep-tech and research labs who may not have access to traditional server infrastructure for AI development purposes and are unable to leverage cluster-scale or cloud AI services due to availability, cost, privacy, and latency concerns.

Super AI Station can be used in a desktop or rack-mounted environment and is delivered as a fully integrated, all-in-one solution including:

Nvidia GB300 Grace Blackwell Ultra Desktop Superchip
Up to 784GB of coherent memory
Integrated Nvidia ConnectX-8 SuperNIC
Closed-loop direct-to-chip liquid cooling for CPU, GPU, ConnectX-8, and memory
Up to 20 PFLOPS AI performance
Bundled with Nvidia AI software stack
Option to configure an additional PCIe GPU for rendering and graphics acceleration
5U desktop tower form factor with optional rack-mounting
1600W power supply compatible with standard power outlets

Rack-Scale GB200 NVL4 GPU Accelerated HPC and AI Solution Now Available

The company is also announcing general availability of its ARS-121GL-NB2B-LCC NVL4 rack-scale platform for GPU-accelerated HPC and AI science workloads such as molecular simulation, weather modeling, fluid dynamics, and genomics. Delivering performance through 4 Nvidia NVLink-connected Blackwell GPUs unified with 2 Nvidia Grace CPUs over NVLink-C2C, up to 32 nodes/rack can be connected via Nvidia ConnectX-8 networking for up to 800G/GPU. The solution is scalable at the system and rack level depending on workload requirements and can be liquid cooled by either in-rack or in-row CDUs.

4xB200 GPUs and 2xGrace Superchips per node with direct-to-chip liquid cooling
4xports of 800G Nvidia Quantum InfiniBand networking per node with 800G dedicated to each B200 GPU (alternative NIC options available)
Up to 128 GPUs in a 48U Nvidia MGX rack for unmatched data center rack density
Power via busbar for seamless scaling

These Supermicro systems are for developing and deploying AI using Nvidia AI Enterprise software and Nvidia Nemotron open AI models.