Super Micro Computer (SMCI) has expanded its artificial intelligence infrastructure portfolio with the introduction of the new H15 server lineup, powered by AMD's latest processors. Alongside the H15 servers, the company launched new GPU systems and rack-scale platforms tailored for enterprise, cloud, and high-performance computing deployments. The announcement deepens SMCI's long-standing partnership with AMD, positioning the company to offer alternatives to NVIDIA-dominated AI server configurations in a market where enterprises and cloud providers are increasingly seeking multi-vendor accelerator options. Following the announcement, SMCI stock closed at $31.03, reflecting a gain of 1.54%, as the update highlighted the company's broader AI infrastructure strategy.

H15 Portfolio Targets Enterprise AI and Cloud Infrastructure

Super Micro Computer introduced its H15 server portfolio utilizing 6th Gen AMD EPYC 9006 Series processors. The newly designed systems support enterprise computing, cloud services, storage, virtualization, and high-performance computing workloads. A key feature of the portfolio is the increase in compute density while maintaining existing power requirements — a design choice that directly addresses growing data center power constraints, where operators face difficulty securing additional electrical capacity for new deployments.

The company engineered the H15 lineup to support up to 256 processor cores and 512 threads. It also expanded memory capacity and input/output bandwidth across the platform. These enhancements allow organizations to efficiently process larger AI workloads and complex enterprise applications.

The H15 family comprises multiple platforms: Hyper, CloudDC, GrandTwin, FlexTwin, Petascale Storage, and SuperBlade. Each platform addresses specific deployment needs across cloud providers, enterprises, research institutions, and hyperscale operators. Furthermore, the systems offer both air-cooled and liquid-cooled configurations to accommodate diverse customer requirements.

New GPU Servers Expand AI Infrastructure Portfolio

Further broadening its AI infrastructure, Super Micro Computer introduced new PCIe GPU servers equipped with AMD Instinct MI350P graphics processors. The AS-5126GS-TNRT and AS-5126GS-TNRT2 models can support up to ten GPUs within a standard 5U air-cooled chassis. These systems deliver higher AI computing capacity without requiring organizations to invest in new cooling infrastructure.

Each server supports AMD Instinct MI350P GPUs featuring up to 144GB of HBM3e memory. The underlying architecture supports low-precision AI computing formats, which are utilized for both inference and training workloads. Consequently, organizations can improve operational performance while simultaneously reducing infrastructure costs and minimizing physical data center space.

To enhance connectivity, the company integrated AMD Pensando Pollara 400 AI network interface cards into its AI infrastructure. This networking platform facilitates front-end communication, storage traffic, and scale-out connectivity using standard Ethernet infrastructure. This approach enables organizations to construct larger AI clusters while maintaining seamless compatibility with existing network environments.

Helios Platform Supports Large-Scale AI Deployments

For expansive AI operations, Super Micro Computer announced the Supermicro AMD Helios Platform. This rack-scale system integrates 72 GPUs featuring AMD Instinct MI455X accelerators alongside 6th Gen AMD EPYC processors. Liquid cooling technology supports higher performance levels while improving operational efficiency across demanding computational workloads.

The Helios Platform also incorporates AMD Pensando networking technologies and the ROCm software stack. ROCm serves as AMD's open-source alternative to NVIDIA's proprietary CUDA platform, and its continued development is central to AMD's effort to expand developer adoption for AI workloads beyond the NVIDIA ecosystem. This combined infrastructure is designed to support large AI training environments and high-throughput inference applications. Organizations have the flexibility to deploy configurations ranging from individual servers to extensive multi-rack installations.

Strategic Integration and Upcoming Showcase

The latest product announcements continue Super Micro Computer's overarching strategy of delivering complete AI infrastructure through its Data Center Building Block Solutions portfolio. The company integrates server hardware, storage, networking, liquid cooling, manufacturing, and deployment services into comprehensive, unified systems. This full-stack integration approach has differentiated SMCI from competitors such as Dell and Hewlett Packard Enterprise in the rapidly growing AI server market.

Super Micro Computer will demonstrate these new platforms during AMD Advancing AI Day 2026 in San Francisco. The demonstration will feature a FlexTwin rack containing 96 AMD EPYC 9006 processors.