Supermicro has introduced its next-generation H15 server portfolio built on AMD’s sixth-generation EPYC 9006 processors, expanding its AI infrastructure lineup with new systems designed for enterprise computing, cloud deployments, high-performance computing and agentic AI workloads.
The new portfolio combines AMD EPYC processors, AMD Instinct GPUs and AMD Pensando networking technologies into a range of server platforms that can scale from individual enterprise deployments to full rack-scale AI infrastructure.
New H15 Servers Target Enterprise AI Growth
The H15 portfolio is designed to address growing demand for AI infrastructure as organizations deploy larger AI models and increasing numbers of autonomous AI agents.
Powered by AMD EPYC 9006 Series processors with up to 256 cores and 512 threads, the servers are intended to support AI inference, enterprise applications, virtualization, cloud computing and high-performance computing while improving compute density and energy efficiency.
Supermicro said the new generation delivers significantly higher CPU performance than its predecessor while increasing memory bandwidth and I/O capacity, allowing organizations to support more concurrent AI workloads without exceeding existing power budgets.
Portfolio Covers Multiple Enterprise Workloads
Rather than offering a single server platform, Supermicro introduced several systems tailored to different deployment scenarios.
The lineup includes:
- Hyper systems for enterprise applications, virtualization and AI inference.
- CloudDC servers designed for hyperscale cloud environments using Open Compute Project standards.
- GrandTwin platforms for scale-out cloud services and object storage.
- FlexTwin liquid-cooled systems built for cloud-native and hyperscale deployments.
- Petascale Storage platforms supporting up to 4.8 petabytes of all-flash storage for AI data lakes and analytics.
- SuperBlade systems optimized for rack-scale AI, HPC and enterprise computing with both CPU and GPU configurations.
The expanded portfolio gives organizations flexibility to deploy infrastructure based on workload requirements while maintaining a common hardware architecture.
AI Infrastructure Expands Beyond Servers
Alongside the H15 systems, Supermicro also announced new GPU-powered AI infrastructure built around AMD Instinct accelerators.
The company introduced new 5U PCIe GPU servers capable of supporting up to 10 AMD Instinct MI350P GPUs in a single air-cooled chassis. The systems are designed for AI inference and training while remaining compatible with existing data center power and cooling environments.
Supermicro also highlighted support for AMD Pensando Pollara 400 AI networking technology, which provides high-bandwidth Ethernet connectivity for large AI clusters and distributed training environments.
Rack-Scale Platform Targets Frontier AI
For organizations building large-scale AI infrastructure, Supermicro is collaborating with AMD on the Supermicro AMD Helios Platform.
The liquid-cooled rack-scale platform combines 72 AMD Instinct MI455X GPUs, sixth-generation AMD EPYC processors, AMD Pensando networking and the AMD ROCm software stack into an integrated AI system designed for frontier AI model training and high-throughput inference.
The company said the platform is intended to support deployments ranging from enterprise AI clusters to large hyperscale environments while improving performance, energy efficiency and operational flexibility.
Focus on Faster AI Deployments
Supermicro says its Data Center Building Block Solutions approach allows customers to deploy validated AI infrastructure more quickly by combining servers, storage, networking, liquid cooling and management technologies into integrated solutions.
The company also plans to demonstrate one of the industry’s highest-density AMD EPYC deployments, featuring 168 EPYC 9006 processors and more than 43,000 CPU cores within a single 48U rack.
The Bottom Line
Supermicro’s H15 portfolio reflects the growing demand for infrastructure capable of supporting enterprise AI at scale. By combining AMD’s latest EPYC processors, Instinct GPUs and Pensando networking into modular server and rack-scale platforms, the company is expanding its portfolio to help organizations deploy AI infrastructure more efficiently while supporting increasingly demanding enterprise and agentic AI workloads.