HPE becomes the first OEM to adopt AMD’s open full-stack AI architecture—”Herder” supercomputer to advance European HPC and Sovereign AI research by 2027
AMD and Hewlett Packard Enterprise are redefining enterprise AI infrastructure. On December 2, 2025, the Tech giant announced an expanded collaboration with HPE to accelerate the next generation of open, scalable AI infrastructure built on AMD leadership compute technologies. HPE will become one of the first system providers to adopt the AMD “Helios” rack-scale AI architecture, which integrates purpose-built HPE Juniper Networking scale-up switches—developed in collaboration with Broadcom—for seamless, high-bandwidth connectivity over Ethernet across massive AI clusters.
Table of Contents
Partnership Highlights
| Category | Details |
|---|---|
| Platform | AMD “Helios” rack-scale AI architecture |
| Performance | Up to 2.9 exaFLOPS FP4 per rack (AMD Instinct MI455X GPUs) |
| Key Components | AMD EPYC “Venice” CPUs, AMD Instinct GPUs, AMD Pensando Vulcano NICs |
| Networking | HPE Juniper switches (Broadcom collaboration), UALoE standard |
| Software Stack | AMD ROCm open software ecosystem |
| Availability | Worldwide rollout in 2026 |
| European Project | “Herder” supercomputer (HLRS, Germany) powered by MI430X GPUs |
| Herder Delivery | Second half 2027 (operational by end 2027) |
Helios: The 2.9 ExaFLOPS Open AI Platform
The AMD “Helios” rack-scale AI platform delivers up to 2.9 exaFLOPS of FP4 performance per rack using AMD Instinct MI455X GPUs, next-generation AMD EPYC “Venice” CPUs, and AMD Pensando Vulcano NICs for scale-out networking, all unified through the open ROCm software ecosystem that enables flexibility and innovation across AI and HPC workloads.
“Helios” combines AMD EPYC CPUs, AMD Instinct GPUs, AMD Pensando advanced networking, and the AMD ROCm open software stack to deliver a cohesive platform optimized for performance, efficiency, and scalability. The system is engineered to simplify deployment of large-scale AI clusters, enabling faster time to solution and greater infrastructure flexibility across research, cloud, and enterprise environments, according to AMD’s official announcement.
Built on the OCP Open Rack Wide design, “Helios” helps customers and partners streamline deployment timelines and deliver a scalable, flexible solution for demanding AI workloads—a critical advantage as enterprises race to deploy multi-thousand GPU clusters for training large language models and generative AI applications.
Why Open Architecture Matters
Dr. Lisa Su, Chair and CEO of AMD, emphasized the strategic significance: “HPE has been an exceptional long-term partner to AMD, working with us to redefine what is possible in high-performance computing. With ‘Helios’, we’re taking that collaboration further, bringing together the full stack of AMD compute technologies and HPE’s system innovation to deliver an open, rack-scale AI platform that drives new levels of efficiency, scalability, and breakthrough performance for our customers in the AI era.”
The “open” designation isn’t marketing fluff—it signals adherence to OCP (Open Compute Project) standards and Ultra Accelerator Link over Ethernet (UALoE), avoiding proprietary lock-in that plagues competitors like NVIDIA’s NVLink. For cloud service providers and enterprises building multi-vendor infrastructure, open standards mean easier integration, reduced vendor dependency, and long-term flexibility. For context on AI infrastructure trends, see TechnoSports’ enterprise technology coverage.
HPE’s Scale-Up Networking Innovation
HPE has integrated differentiated technologies for customers, specifically a scale-up Ethernet switch and software designed for “Helios.” Developed in collaboration with Broadcom, the switch delivers optimized performance for AI workloads using the Ultra Accelerator Link over Ethernet (UALoE) standard, reinforcing AMD’s commitment to open, standards-based technologies.
Antonio Neri, President and CEO at HPE, positioned this as infrastructure evolution: “For more than a decade, HPE and AMD have pushed the boundaries of supercomputing, delivering multiple exascale-class systems and championing open standards that accelerate innovation. With the introduction of the new AMD ‘Helios’ and our purpose-built HPE scale-up networking solution, we are providing our cloud service provider customers with faster deployments, greater flexibility, and reduced risk in how they scale AI computing in their businesses.”
The networking layer is critical—AI training clusters require ultra-low-latency, high-bandwidth interconnects to synchronize gradients across thousands of GPUs. Traditional Ethernet couldn’t match NVIDIA’s proprietary InfiniBand/NVLink, but UALoE closes this gap through hardware offloading and protocol optimization, making Ethernet viable for exascale AI workloads.
“Herder” Supercomputer: Sovereign AI for Europe
Herder, a new supercomputer for the High-Performance Computing Center Stuttgart (HLRS) in Germany, is powered by AMD Instinct MI430X GPUs and next-generation AMD EPYC “Venice” CPUs. Built on the HPE Cray Supercomputing GX5000 platform, Herder will offer world-class performance and efficiency for HPC and AI workloads at scale. The combination of AMD’s leadership compute portfolio with HPE’s proven system design will create a powerful new tool for sovereign scientific discovery and industrial innovation for European researchers and enterprises.
Prof. Dr. Michael Resch, Director of HLRS, explained the strategic rationale: “The pairing of AMD Instinct MI430X GPUs and EPYC processors within HPE’s GX5000 platform is a perfect solution for us at HLRS. Our scientific user community requires that we continue to support traditional applications of HPC for numerical simulation. At the same time, we are seeing growing interest in machine learning and artificial intelligence. Herder’s system architecture will enable us to support both of these approaches, while also giving our users the ability to develop and benefit from new kinds of hybrid HPC/AI workflows.”
Delivery of Herder is scheduled for the second half of 2027 and expected to go into service by end of 2027. Herder will replace HLRS’s current flagship supercomputer, called Hunter—marking a generational leap in European computational capacity at a time when AI sovereignty concerns drive governments to build domestic infrastructure independent of US cloud giants.
Europe’s AI Sovereignty Push
The Herder project reflects broader European priorities: maintaining technological independence through domestically controlled compute infrastructure. As EU regulations like the AI Act demand local data processing and transparency, supercomputers like Herder enable European researchers and enterprises to train AI models without relying on AWS, Azure, or Google Cloud—addressing both regulatory compliance and geopolitical risk, according to European HPC strategy documents.
Professor Resch added: “This platform will not only make it possible for our users to run larger, more powerful simulations that lead to exciting scientific discoveries, but also to develop more efficient computational methods that are only feasible with the capabilities that such next-generation hardware offers.”
Market Positioning: AMD vs. NVIDIA
AMD’s Helios directly challenges NVIDIA’s H200/B200 GPU clusters that currently dominate enterprise AI infrastructure. Key differentiators:
- Open standards (OCP, UALoE) vs. NVIDIA’s proprietary NVLink/InfiniBand
- Price competitiveness (AMD typically 20-30% cheaper than NVIDIA equivalents)
- Software maturity gap (ROCm improving but still behind CUDA ecosystem)
HPE’s adoption as the first major OEM validates AMD’s enterprise credibility. If cloud providers like Microsoft Azure or Oracle Cloud follow HPE’s lead, AMD could capture 15-25% of the AI accelerator market by 2027—still trailing NVIDIA’s 80%+ dominance but establishing a viable alternative for cost-conscious buyers and open-standards advocates.
What’s Next
HPE will offer the AMD “Helios” AI Rack-Scale Architecture worldwide in 2026, with early deployments likely targeting hyperscalers, national labs, and enterprises building private AI clouds. The 2027 Herder delivery will provide real-world performance benchmarks—critical for convincing skeptical buyers that AMD can deliver exascale AI infrastructure at production scale.
For AMD, success hinges on ROCm software maturation (PyTorch/TensorFlow optimization), ecosystem partnerships (ISVs porting models to ROCm), and sustained price advantages as NVIDIA’s Blackwell generation launches.
Also Read: AI Infrastructure Trends 2025 | Europe’s Sovereign AI Strategy
