Hewlett Packard Enterprise (NYSE: HPE) announced the HPE ProLiant Compute DL394 Gen12 server powered by NVIDIA Vera CPU, designed for agentic AI and reinforcement learning applications. The company made the announcement at COMPUTEX, stating the server will be available in fall 2026.

The 2U server features NVIDIA Vera CPUs with a monolithic design that addresses non-uniform memory access issues found in traditional chiplet architectures. Using LPDDR5X memory technology, the server achieves 1.2 TB/s aggregate bandwidth and up to 14 GB/s per core, according to the company.

"The shift from generative models to agentic systems is redefining the role of compute across the enterprise," said Antonio Neri, president and CEO of HPE. "These workloads require high-performance servers with exceptional CPU performance to enable real-time reasoning across agentic AI and financial services applications."

The New York Stock Exchange is exploring the technology in collaboration with HPE, NVIDIA, and Redpanda. "NYSE processes more than 1.1 trillion messages per day, and in collaboration with Redpanda and HPE, using NVIDIA Vera CPUs, we will be scaling our capacity while further optimizing latency," said Lynn Martin, President of NYSE Group.

NVIDIA CEO Jensen Huang stated that "Vera was built to orchestrate AI factories—delivering 2x the efficiency and faster task completion than x86."

The server includes HPE's Silicon Root of Trust firmware technology and iLO 7 management capabilities. HPE states the servers meet NIST's quantum computing resistant security requirements. The system also features HPE Compute Ops Management for unified server environment management.

The server will be part of the NVIDIA AI Computing by HPE portfolio and can be acquired through HPE Financial Service's 90/9 Advantage program, which offers no payments for 90 days and nine months at one percent interest.