Supermicro expands its portfolio of new GPU systems based on NVIDIA reference architectures, based on NVIDIA architecture.

Super Micro Computer, Inc. (SMCI) today announced the availability of servers based on the NVIDIA GH200 Superchip, the industry’s first NVIDIA MGX family of products.

The new modular architecture standardizes AI infrastructure and accelerated computing in compact 1U and 2U form factors, while providing flexibility and scalability for GPUs, DPUs, and CPUs. Supermicro’s high-performance liquid cooling technology enables dense configurations, such as a 1U two-node build with two NVIDIA GH200 Grace Hopper superchips integrated with a high-speed interconnect. Additionally, Supermicro has the ability to supply thousands of rack-scale AI servers per month from its global facilities, ensuring plug-and-play compatibility.

“Supermicro is recognized as a leader in AI innovation today, transforming the data center to realize the potential of AI across diverse workloads,” said Charles Liang, president and CEO of Supermicro. “It is critical that we deliver highly modular, scalable and versatile systems for the rapid evolution of AI technologies. Supermicro’s NVIDIA MGX-based solutions demonstrate how our building block solutions strategy can help bring the latest systems to market quickly and are the industry’s most workload-optimized. Supermicro is working with NVIDIA to help enterprises accelerate time to market for new AI-based applications while simplifying deployment and reducing environmental impact. The new server families include the latest industry-leading AI-optimized technologies, including the NVIDIA GH200 Grace Hopper superchip, BlueField, and PCIe 5.0 EDSFF slots,” he added.

“NVIDIA has a long history of working with Supermicro to deliver the highest-performance AI systems,” said Ian Buck, vice president of Hyperscale and HPC at NVIDIA. “The NVIDIA MGX modular reference design, along with Supermicro’s server expertise and NVIDIA Grace Hopper superchips, will enable next-generation AI systems that will benefit a broad range of customers around the world.”

Supermicro’s NVIDIA MGX platform will offer a wide range of servers capable of accommodating next-generation AI technologies. Key new products include: ARS-111GL-NHR ARS-111GL-NHR-LCC ARS-111GL-DHNR-LCC ARS-121L-DNR ARS-221GL-NR SYS-221GE-NR.

All MGX platforms can be enhanced with NVIDIA BlueField-3 DPU and/or NVIDIA ConnectX-7 interconnect for high-performance InfiniBand or Ethernet networking.