Back
Product News
MSI Showcases Advanced AI Infrastructure from Rack Scale to the Desktop at SC26
November 18, 2026
AIoT Solutions
Liquid-cooled platforms based on NVIDIA MGX and desktop AI supercomputing advance scalable enterprise AI infrastructure

Chicago, Illinois – November 17, 2026 – MSI, a leading global provider of high-performance server solutions, showcases its AI infrastructure portfolio at SuperComputing 2026 (SC26), booth #5429. Spanning rack-scale accelerated computing, flexible AI computing platforms, and desktop AI supercomputing, MSI combines NVIDIA technologies with its system integration expertise to help enterprises build scalable AI infrastructure, accelerate AI development, and support the next generation of AI applications.
“As AI workloads continue to grow in scale and complexity, enterprises need infrastructure that can adapt to evolving requirements,” said Danny Hsu, General Manager of Enterprise Platform Solutions at MSI. “Through our collaboration with NVIDIA, we combine advanced computing technologies with MSI’s system integration expertise to help customers scale AI deployments, optimize infrastructure investments, and build a flexible foundation for future innovation.”
Build Rack-Scale AI Infrastructure with NVIDIA MGX
MSI showcases an NVIDIA MGX portfolio, spanning integrated rack-scale infrastructure and AI computing platforms, with flexible deployment options for demanding enterprise AI and accelerated computing workloads. NVIDIA MGX is a modular reference architecture for accelerated computing, helping system builders design future-compatible systems faster, reducing costs, accelerating time to market, and improving ROI.
MSI’s 48RU rack based on NVIDIA MGX architecture integrates four CG681-S6093 6U liquid-cooled GPU servers with NVIDIA Spectrum-4 SN5600 data switches and SN2201 out-of-band networking switches. By bringing computing, high-speed networking, and liquid cooling together at rack scale, the solution provides an integrated foundation for high-density accelerated computing and scalable AI Factory deployments.
Each CG681-S6093 supports dual AMD EPYC™ 9006 Series Server processors and up to eight liquid cooled NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs. The platform also supports up to eight NVIDIA ConnectX-8 SuperNICs with 400Gbps Ethernet connectivity per port, PCIe 6.0 expansion, and up to 32 DDR5 DIMM slots, delivering a high-density computing foundation for demanding AI and accelerated computing workloads.
Platforms based on NVIDIA MGX for Flexible AI Deployments
Beyond rack-scale infrastructure, MSI offers platforms based on NVIDIA MGX architecture in multiple form factors to accommodate different performance, GPU density, and deployment requirements.
The CG480-S6053 is a 4U platform powered by dual AMD EPYC™ 9005 Series processors. It supports up to eight double-width GPUs, 24 DDR5 DIMM slots, eight U.2 NVMe drives, and five PCIe 5.0 expansion slots, providing flexibility for enterprise AI and accelerated computing configurations.
For deployments where rack space is a priority, the CG290-S3063 is a compact 2U platform powered by Intel® Xeon® 6 processors. It supports up to four double-width GPUs, 16 DDR5 DIMM slots, and four rear U.2 NVMe drives, offering a flexible option for AI inference and other space-conscious data center workloads.
Together, MSI’s solutions based on NVIDIA MGX provide options ranging from integrated rack-scale infrastructure to standalone systems, helping enterprises align their infrastructure with evolving AI workloads and data center requirements.
Accelerate Enterprise AI Development with MSI XpertStation WS300
For enterprise AI teams seeking powerful local computing for AI development, the MSI XpertStation WS300, built on the NVIDIA DGX Station platform , brings desktop AI supercomputing to the workplace. Designed for AI development, fine-tuning, inference, and data-intensive workflows, it will support Windows for local AI application and agent development.
Powered by the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip, the WS300 delivers up to 748GB of coherent memory with ECC and 7.1TB/s of HBM3e GPU memory bandwidth to accelerate CPU-GPU communication and large-model processing. Dual 400GbE networking via NVIDIA ConnectX-8 SuperNICs, combined with a liquid-cooled thermal design, brings data-center-class AI computing to a compact deskside form factor.