Supermicro introduces new server portfolio
Super Micro Computer has announced its next-generation H15 server portfolio powered by 6th Gen AMD EPYC 9006 Series CPUs, optimised for next generation GPUs including AMD Instinct, and connected by AMD Pensando networking. With up to 256 cores and 512 threads, H15 systems meet the growing compute demands of cloud, enterprise, storage, high-performance computing (HPC), and agentic AI workloads. With breakthrough 1.7x generational CPU performance improvement, expanded memory and I/O bandwidth, and industry-leading compute density, customers are now able to run more concurrent AI agents, accelerate enterprise applications, and maximise host-node performance while operating within existing power envelopes.
The new H15 portfolio includes purpose-built systems optimised for a broad range of enterprise and AI infrastructure deployments:
Hyper – The flagship dual-socket platform engineered for enterprise applications, AI inference, virtualisation, and cloud workloads, with advanced thermal design to support the highest-performance AMD EPYC processors.
CloudDC – A single-socket or dual socket server designed for cloud-scale environments and built on the Open Compute Project (OCP) Data Center Modular Hardware System (DC-MHS) specification for compatibility with open data centre standards.
GrandTwin – A high-density 2U, four-node architecture designed for scale-out environments including object storage, virtualisation, cloud services, and high-performance computing.
FlexTwin– A 1OU, two-node, high-performance, high-density dual CPU compute system that maximises compute density and power efficiency for cloud-native and hyperscale deployments using liquid cooling.
Petascale Storage – 1U and 2U high-capacity all-flash storage platforms optimised for software-defined storage-based AI data lakes, large-scale analytics, and HPC environments supporting up to 4.8 PB per system.
SuperBlade – H15 8U 10 SuperBlade represents next-generation, rack-scale breakthrough architecture for HPC, AI inference, agentic AI, and enterprise-class compute workloads with CPU and GPU. The platform supports both single-socket and dual-socket blade configurations – with both air and liquid-cooled versions optimised for maximum density, high performance, and efficiency, across a wide range of infrastructure deployments.
Complementing the H15 server portfolio, Supermicro continues to expand its AMD GPU-powered AI infrastructure with new PCIe GPU servers and the rack-scale Supermicro AMD Helios Platform. As shown at Computex 2026, these solutions provide organisations with flexible deployment options ranging from enterprise AI inference to large-scale AI training.
5U PCIe GPU Servers Powered by AMD Instinct MI350P GPUs
The Supermicro AS -5126GS-TNRT and AS -5126GS-TNRT2 are designed to maximise the performance of AMD Instinct MI350P PCIe GPUs. Supporting up to ten GPUs in a standard 5U air-cooled platform, these systems deliver exceptional AI acceleration while operating within existing data centre power and cooling infrastructures.
By combining Supermicro’s high-density PCIe architecture with AMD Instinct MI350P GPUs featuring up to 144GB of HBM3e memory and support for low-precision AI formats, organizations can accelerate AI inference and training while improving infrastructure efficiency, reducing data centre footprint, and lowering total cost of ownership.
Open Ethernet Networking with AMD Pensando Pollara 400 AI NIC
The AMD Pensando Pollara 400 AI NIC provides high-performance, open Ethernet networking for AI infrastructure, enabling front-end, storage, and scale-out connectivity for AMD Instinct MI350P-based systems with the high bandwidth, low latency, and efficiency required for AI training and inference. Together, AMD Instinct MI350P GPUs and the AMD Pensando Pollara 400 AI NIC enable customers to build open, high-performance AI clusters that scale from a single server to large multi-rack deployments using standard Ethernet infrastructure.
Supermicro AMD Helios Platform
For organisations deploying frontier AI models, Supermicro is collaborating with AMD to deliver the Supermicro AMD Helios Platform, a 72-GPU rack-scale solution designed for large-scale AI training and high-throughput inference.
The liquid-cooled platform combines AMD Instinct MI455X GPUs, 6th Gen AMD EPYC processors, AMD Pensando networking technologies, and the AMD ROCm software stack to create an open, high-performance AI infrastructure. Supporting deployments of every size, the platform enables customers to scale efficiently while maximising performance, energy efficiency, and operational flexibility.
Supermicro’s DCBBS brings these technologies together as complete, validated AI infrastructure, enabling organisations to deploy solutions ranging from individual servers to fully integrated rack-scale and data centre-level systems.
