MSI showcased cloud and AI-optimized server platforms at CloudFest 2026.

Cloud infrastructure is evolving rapidly as AI services and data-intensive workloads continue to expand, according to Danny Hsu, General Manager of MSI's Enterprise Platform Solutions. "Service providers and enterprises need platforms that can scale efficiently while maintaining performance and operational flexibility across next-generation data center environments," Hsu stated. MSI's portfolio addresses these demands through platforms designed for hyperscale and service provider deployments, offering deployment flexibility across 2U 4-node, 2U 2-node, and ORv3 21-inch server architectures powered by AMD EPYC and Intel Xeon 6 processors. Built on DC-MHS (Data Center Modular Hardware System) architecture, the platforms improve serviceability and hardware interoperability, allowing operators to streamline maintenance and reduce operational complexity. For hyperscale environments, MSI provides an ORv3-compliant 21-inch platform designed for Open Compute racks and 48V rack power architectures, enabling improved compute density, power efficiency, and seamless rack-scale integration.

The CD270-S3071-X4 exemplifies the company's scalable compute offering, a 2U 4-node platform designed for high-density cloud deployments. Each node supports a single Intel Xeon 6 processor with up to 400W TDP, 12 DDR5 DIMM slots across 12 memory channels, and 3 front U.2 NVMe bays, enabling high compute density and efficient scaling for cloud-scale workloads.

MSI's AI platforms extend the portfolio from large-scale training clusters to local development environments with NVIDIA-accelerated systems combining NVIDIA GPUs, high-bandwidth memory architectures, PCIe 5.0/6.0 connectivity, and high-speed networking. The CG481-S6053 is a 4U NVIDIA MGX server powered by dual AMD EPYC 9005 processors, supporting up to 8 NVIDIA RTX PRO 6000 Blackwell Server Edition or NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs and 8 400Gb QSFP112 networking ports, optimized for large-scale AI training clusters and model fine-tuning. The CG290-S3063 is a 2U NVIDIA MGX server built on Intel Xeon 6 processors supporting up to 4 NVIDIA RTX PRO 6000 Blackwell Server Edition or NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs, well suited for scalable inference and distributed AI deployments. The XpertStation WS300, based on NVIDIA DGX Station architecture, integrates an NVIDIA Grace CPU paired with an NVIDIA Blackwell Ultra GPU, featuring a 748GB coherent memory pool and 2×400GbE networking powered by NVIDIA ConnectX-8 SuperNICs, enabling developers to develop, prototype, fine-tune, and validate AI models locally before production deployment.

For hybrid cloud environments, MSI offers enterprise platforms spanning 1U and 2U systems powered by AMD EPYC 9005 and Intel Xeon 6 processors. The CX270-S5062 is a 2U dual-socket Intel Xeon 6 platform supporting 32 DDR5 DIMMs, up to 24 front U.2 NVMe drives, and 2 PCIe 5.0 double-wide GPUs. The CX271-S4056 (-HE SKU) is a 2U single-socket AMD EPYC 9005 platform with 24 DDR5 DIMMs, 2 PCIe 5.0 double-wide GPUs, and NVMe storage. The CX171-S4056 is a 1U AMD EPYC 9005 platform supporting 24 DDR5 DIMMs and up to 12 front NVMe drives, enabling high compute density for space-efficient data center deployments.