Models & Labs · United States

HPE’s Gen13 Signals Enterprise AI Shift to CPU-GPU Balance

Credited to Data Center Knowledge · datacenterknowledge.com · 4h ago

Photograph · Data Center Knowledge

HPE launched AMD Epyc-powered ProLiant Gen13 servers for enterprise AI inference, agentic AI, and traditional workloads, targeting enterprise customers with varied AI needs.

HPE today (Oct. 7) announced the first batch of next-generation, AMD-powered servers designed to enable enterprises to run AI inference and agentic AI, as well as traditional enterprise workloads such as analytics and virtualization.

The company unveiled its first four ProLiant Gen13 servers, all powered by AMD’s sixth-generation Epyc 9006 SP7-series CPUs – code-named “Venice” – featuring new automated security and expanded management capabilities.

The new high-end ProLiant DL585a Gen13 server, which features up to eight GPUs and two AMD Venice processors, is designed for demanding enterprise workloads such as AI inference, agentic AI, retrieval-augmented generation (RAG) model fine-tuning, and analytics, the company said. The server, available in March 2027, will support GPUs from Nvidia, AMD, and Intel, said John Carter, HPE’s vice president of server product management, quality, and technical pursuit.

Related: HPE Interview: Why Data Center Efficiency Is Now Core to IT Decisions

“These are not the big [Nvidia Vera Rubin] NVL72 or AMD Helios scale-up type systems. This is for your enterprise-grade customer,” Carter said in an interview with Data Center Knowledge. “This is the first time we’ve done this kind of enterprise-grade AI system with an AMD head node. There is a lot of great performance improvements there, utilizing the larger memory bandwidth, higher-performance CPU.”

HPE also introduced three CPU-focused servers. The compact ProLiant DL525 server, available in November 2026, is a 1U, single-socket system featuring one AMD Venice chip with up to 256 cores, designed for AI inference and traditional enterprise workloads, such as virtualization, fraud detection, market data processing, modeling, and simulations, the company said.

The other two new servers, available in 2027, are CPU-based, rack-scale systems that offer high-density compute for very large enterprises or service providers, such as specialist cloud providers (often called neoclouds ) that have data-intensive AI or HPC workloads, Carter said. The ProLiant XD245 is a four-node, liquid-cooled system, while the ProLiant XD285 is a two-node, air-cooled system.

Analysts: Epyc-Powered ProLiants Aim at Enterprise AI

HPE’s server announcement signals that the company sees enterprise AI adoption on the horizon and wants to be aggressive in meeting enterprise needs, said analyst Matt Kimball of Moor Insights & Strategy. AMD’s sixth-generation Epyc chip is a strong processor for AI because of its high core count, memory capacity, and bandwidth, he added.

IDC research vice president Kuba Stolarski agreed, saying HPE’s four new servers provide organizations with a good mix of hardware for their workload needs.

Related: HPE Targets GPU Utilization With New AI Networking Portfolio

“Every enterprise is going to have a variety of different AI use cases. So you can’t have a one-size-fits-all from an infrastructure perspective,” Stolarski told Data Center Knowledge. “You can’t just put up the largest, most expensive GPU cluster and expect that’s going to be the best solution for every enterprise.”

How ProLiant Gen13 Systems Support Agentic AI Workflows

HPE executives and analysts said the three new CPU-focused ProLiant Gen13 servers can work alongside GPU-based servers for agentic AI workflows. For example, the CPU-focused servers can serve as a CPU head node in an agentic AI system, handling orchestration, tool calls, and retrieval, while GPU systems handle inference.

“You need a whole lot more CPUs to drive orchestration because these agents are making calls, and subagents are making calls. It’s just an ongoing traffic frenzy, and that’s what makes AMD’s Venice so interesting,” Kimball said. “That high core count and high single-threaded performance do a very good job of supporting orchestration for agentic AI.”

Security and Management: iLO 8 Automation and PQC Readiness

Chris Bradley, director of mainstream compute customer advocacy and technical enablement at HPE, said the ProLiant Gen13 systems will feature new automated security features as part of HPE Integrated Lights-Out 8, including support for self-encrypting drives and expanded post-quantum cryptography readiness.

Related: Discover 2026: HPE Bets on Hybrid Quantum-Supercomputing Architectures

New features in HPE Compute Ops Management software include low-touch onboarding and near-real-time power capacity insights, he said.

About the Author

Wylie Wong is a journalist and freelance writer specializing in technology, business and sports. He previously worked at CNET, Computerworld and CRN and loves covering and learning about the advances and ever-changing dynamics of the technology industry. On the sports front, Wylie is co-author of Giants: Where Have You Gone, a where-are-they-now book on former San Francisco Giants. He previously launched and wrote a Giants blog for the San Jose Mercury News, and in recent years, has enjoyed writing about the intersection of technology and sports.

Want more Data Center Knowledge stories in your Google search results?

Editor's Choice

Industry Voices

Join the entire data center ecosystem September 21 – 23, 2026, at the Gaylord Texan Resort & Convention Center in Dallas, TX.

Power & SitesUnited States

Doosan Enerbility Ships 380-MW Gas Turbine to SpaceXAI

Credited to TheElec · thelec.net1d ago

Useful

Doosan Enerbility has completed production of a 380-megawatt (MW) gas turbine to be supplied to US artificial intelligence (AI) company SpaceXAI.

Doosan Enerbility held a shipment ceremony at its Changwon headquarters for the S2 380-MW gas turbine (DGT6-300H). About 70 people attended the event, including SpaceXAI officials and Son Seung-woo, head of Doosan Enerbility's Power Service Business Group.

The gas turbine will be deployed in an energy infrastructure project being developed by SpaceXAI in the US. Doosan Enerbility manufactured the turbine to meet the technical specifications requested by SpaceXAI.

"By entering the supply chain of a global AI company, we have demonstrated the technology and quality of Korean gas turbines," Son said. "We will expand our presence in the global large gas turbine market, including North America."

Original · TheElec

Front Method

Models & Labs · United States

Microchip, Navitas Unveil 800V Power Reference Design for AI Servers

Credited to TheElec · thelec.net · 1d ago

Microchip Technology and Navitas Semiconductor have unveiled a power-conversion reference design for artificial intelligence (AI) data centers. The design converts 800V direct current supplied to a server rack directly to 6V inside the server.

Microchip Technology and Navitas Semiconductor announced Oct. 6 that they introduced the power-conversion reference design, which combines two stages into one to reduce conversion losses compared to conventional two-stage conversion designs that step 800V down to 50V and then convert it again to 6V. The 6V output is ultimately converted to less than 1V by a voltage regulator module (VRM) on the graphics processing unit (GPU) board.

The board uses 16 gallium nitride (GaN) power semiconductors from Navitas. GaN can switch electricity on and off faster than silicon while generating less loss in the process, allowing power-conversion components to be made smaller while maintaining efficiency. The board is about 20% thinner than a smartphone, allowing it to be positioned close to the GPU board.

Microchip supplies the digital signal controller (DSC) dsPIC33AK, which controls power conversion, and the TA100 security chip. The TA100 handles functions including secure boot and firmware authentication.

Navitas unveiled the board's basic power architecture at Nvidia's GTC 2026 developer conference in March. This time, the companies combined Microchip's controller and security chip to offer a joint reference design. A reference design is a baseline design provided for server and power-equipment companies to use in product development. The two companies are providing the board, software, and design documentation together.

AI data centers are seeing a sharp increase in power consumption per rack as large numbers of GPUs are deployed. Nvidia is building an ecosystem around designs that deliver 800V direct current all the way to the rack, an approach that can reduce power losses and the scale of wiring required.

Nvidia expects the approach to gain wider adoption starting with shipments of its next-generation rack system, "Kyber," in 2027.

​Doosan Enerbility Ships 380-MW Gas Turbine to SpaceXAI

Doosan Enerbility Ships 380-MW Gas Turbine to SpaceXAI