Supermicro has unveiled a next-generation H15 server portfolio equipped with 6th generation AMD EPYC 9006 series CPUs. The new lineup features up to 1.7 times better CPU performance compared to the previous generation and is configured to cover cloud, enterprise, and AI inference and training environments.

Supermicro announced the launch of its H15 server portfolio on the 27th.

It was announced that the H15 system supports up to 256 cores and 512 threads, and has enhanced high-performance workload processing capabilities through a 33% increase in core count, double the PCIe bandwidth, and 2.6 times the memory bandwidth.

It works with AMD Instinct GPUs and AMD Pensando networking, and is based on the Data Center Building Blocks Solution (DCBBS) architecture.

Vik Malyala, Chief Business Officer at Supermicro, stated, “The new AMD-based DCBBS product family provides next-generation AI infrastructure featuring high performance, rapid scalability, and high efficiency,” adding, “We are supporting customers in adopting and expanding their AI infrastructure through our global service team and U.S. supply chain.”

"Combining the latest AMD EPYC CPUs, Instinct GPUs, and AMD Pensando networking with Supermicro's modular design enables the rapid deployment of AI infrastructure while increasing system utilization and reducing energy consumption and total cost of ownership (TCO)," said Dan McNamara, Senior Vice President of AMD's Compute and Enterprise AI business unit.

The H15 portfolio consists of six types of systems classified by use.

△Hyper: Dual-socket platform for enterprise, AI inference, and virtualization △CloudDC: Single and dual-socket servers based on OCP DC-MHS standards △GrandTwin: 2U 4-node high-density for scale-out environments Architecture △FlexTwin™: A liquid-cooled 1U 2-node hyperscale system △Petascale Storage: An all-flash storage platform supporting up to 4.8 PB per system △SuperBlade: An 8U 10-node rack-scale architecture for HPC, AI inference, and agentic AI.

Supermicro has expanded its AMD GPU-based AI infrastructure lineup by adding PCIe GPU servers and rack-scale AMD Helios platforms along with the H15 portfolio.

It was announced that the 5U server (AS-5126GS-TNRT·AS-5126GS-TNRT2) equipped with AMD Instinct MI350P PCIe GPU supports up to 10 GPUs and 144GB of HBM3e memory, and can be operated while maintaining existing data center power and cooling infrastructure.

The AMD Pensando Polara 400 AI NIC provides high-bandwidth, low-latency Ethernet networking required for AI training and inference, and supports front-end, storage, and scale-out connectivity for MI350P-based systems.

The Supermicro AMD Helios platform is a 72-GPU rack-scale liquid-cooled solution integrating AMD Instinct MI455X GPUs, 6th Gen AMD EPYC processors, AMD Pensando networking, and AMD ROCm software, targeting large-scale AI training and high-throughput inference environments.