HPE Introduces CPU Server with NVIDIA Vera CPU, Purpose-built for Agentic AI
At COMPUTEX 2026 in Taipei, Hewlett Packard Enterprise made a bold declaration about where enterprise computing is heading. The company unveiled the HPE ProLiant Compute DL394 Gen12—a purpose-built server powered by the NVIDIA Vera CPU, engineered from the ground up to meet the relentless performance demands of agentic AI, reinforcement learning, and AI-factory-scale data processing.
The announcement is more than a hardware refresh. It signals a fundamental shift in how enterprise infrastructure must be conceived as organizations move beyond generative AI into a new era of autonomous, reasoning systems. Agentic AI workloads—where AI agents plan, reason, and execute tasks with minimal human intervention—demand a different class of compute: lower latency, higher memory bandwidth, and deterministic performance that conventional server architectures simply cannot reliably provide.
The HPE ProLiant Compute DL394 Gen12 was unveiled as part of a landmark three-way collaboration with NVIDIA and data streaming company Redpanda, with the New York Stock Exchange (NYSE) already exploring this infrastructure to power one of the world's most demanding real-time data environments.
Why Agentic AI Demands a New CPU
Traditional server architectures built around high-core-count x86 chiplet designs were not designed with today's AI workloads in mind. Multi-chip configurations introduce non-uniform memory access (NUMA) issues, creating variable latencies and non-deterministic performance that can critically undermine the precision and speed required by agentic AI agents operating in real time.
The NVIDIA Vera CPU addresses this with a monolithic architecture that eliminates NUMA inefficiencies. By pairing this design with LPDDR5X memory—a highly efficient form of dynamic random-access memory—the DL394 Gen12 achieves up to 14 GB/s per core, enabling it to ingest and process data at speeds that match the cadence of autonomous AI reasoning. The result is an architecture where the NVIDIA Vera CPU acts as a high-speed orchestrator, efficiently balancing resources across the most demanding AI workloads while reducing capacity waste.
"The shift from generative models to agentic systems is redefining the role of compute across the enterprise. These workloads require high-performance servers with exceptional CPU performance to enable real-time reasoning across agentic AI and financial services applications. With our new HPE ProLiant Compute server, we are delivering a new class of infrastructure to help customers accelerate insights and operate with confidence in the most demanding environments."
— Antonio Neri, President & CEO, HPEInside the HPE ProLiant Compute DL394 Gen12
The DL394 Gen12 is a 2U server that packages extraordinary compute density into a compact, data-center-ready form factor. Beyond raw performance, HPE has engineered enterprise-grade security and management directly into the platform's foundations—a deliberate design choice for organizations deploying AI in regulated, high-stakes environments.
NVIDIA Vera CPU with LPDDR5X Memory
The monolithic NVIDIA Vera CPU design delivers 1.2 TB/s aggregate memory bandwidth, up to 14 GB/s per core. Unlike chiplet-based x86 architectures that introduce NUMA latency, Vera's unified design ensures deterministic, low-latency performance—critical for agentic AI tasks requiring real-time reasoning and sequential logic execution.
HPE Integrated Lights-Out (iLO 7) Security
Security is embedded at the silicon level via HPE's Silicon Root of Trust firmware technology. The iLO 7 implementation introduces a secure enclave—an HPE innovation that protects servers throughout their entire lifecycle. Notably, the DL394 Gen12 is among the first servers to meet NIST's quantum-computing-resistant security requirements, future-proofing sensitive workloads against next-generation threats.
HPE Compute Ops Management
A unified, AI-driven management layer that provides a single dashboard for overseeing distributed server environments. By automating routine management tasks, Compute Ops Management reduces the time required to manage server fleets and minimizes the risk of revenue-impacting downtime—enabling IT teams to focus on strategic AI deployment rather than operational firefighting.
Agentic AI Data Storage & Processing Stack
The server's optimized technology stack spans data storage, processing, monitoring, management, and security—each component tuned for the specific demands of agentic AI operations. Through the collaboration with Redpanda for high-throughput data streaming, the platform provides the full-stack infrastructure required by agentic systems operating at scale.
NVIDIA's Vision: A New CPU for the AI Factory Era
Jensen Huang's framing of the Vera CPU at COMPUTEX was unambiguous: the transition to agentic AI is not an incremental evolution of existing infrastructure—it is a categorical shift requiring entirely new silicon. The Vera CPU, built to orchestrate AI factories, is designed to deliver twice the efficiency of x86 alternatives and faster task completion for workloads that demand deterministic, high-bandwidth compute.
"Agentic AI has arrived, and it needs a new CPU. Vera was built to orchestrate AI factories—delivering 2x the efficiency and faster task completion than x86. With HPE ProLiant Compute DL394 Gen12, enterprises can put Vera to work, and NYSE shows what purpose-built AI infrastructure can do in the world's most demanding environments."
— Jensen Huang, Founder & CEO, NVIDIANYSE: A Real-World Stress Test at 1.1 Trillion Messages Per Day
Few environments in the world present a more demanding test for infrastructure than a major financial exchange. The New York Stock Exchange processes over 1.1 trillion messages per day—a relentless torrent of market data, trade signals, and system communications where microseconds of latency can translate directly into financial impact.
NYSE's exploration of the HPE ProLiant Compute DL394 Gen12—in collaboration with HPE, NVIDIA, and Redpanda—positions the platform as a credible solution for the most mission-critical financial infrastructure. For enterprise architects evaluating the server, the NYSE use case provides a compelling proof point: if the platform can meet the throughput, latency, and resilience demands of a global stock exchange, it can handle virtually any enterprise AI workload.
HPE has launched the ProLiant Compute DL394 Gen12 at COMPUTEX 2026—a 2U server powered by the NVIDIA Vera CPU—specifically engineered for agentic AI, reinforcement learning, and AI-factory-scale data processing workloads.
The server's monolithic NVIDIA Vera CPU design—paired with LPDDR5X memory—eliminates NUMA latency issues common in chiplet x86 architectures, delivering 1.2 TB/s aggregate memory bandwidth and up to 14 GB/s per core.
Enterprise-grade security is embedded at every layer via HPE's Silicon Root of Trust, iLO 7 secure enclave, and NIST quantum-computing-resistant protections—making the DL394 Gen12 one of the first servers to meet next-generation cryptographic standards.
A three-way collaboration between HPE, NVIDIA, and Redpanda underpins the platform's data streaming capabilities, with the New York Stock Exchange actively exploring the solution for its infrastructure processing over 1.1 trillion messages per day.
The DL394 Gen12 is planned for fall 2026 availability, positioning it as a key infrastructure option for enterprises currently planning their transition from generative AI to agentic AI deployments at scale.
With the HPE ProLiant Compute DL394 Gen12 slated for fall 2026 availability, enterprise technology teams now have a concrete timeline for evaluating this platform in their AI infrastructure roadmaps. For organizations that have been waiting for purpose-built agentic AI infrastructure—rather than adapting GPU-centric or legacy x86 servers for workloads they were never designed to handle—this announcement marks the arrival of a genuine alternative. The NYSE collaboration provides an industry-validated proof point that is difficult to overlook.
As the enterprise compute landscape continues its rapid transformation, HPE's collaboration with NVIDIA and Redpanda positions the ProLiant Compute DL394 Gen12 as a foundational building block for the agentic AI era. To explore the full HPE server portfolio and learn more about the ProLiant Compute DL394 Gen12, visit www.hpe.com.