Skip to main content
HPE AI Factory and NVIDIA reveal Vera, the world’s first AI agent-optimized CPU, showcasing next-gen AI infrastructure innova

Editorial illustration for HPE AI Factory and NVIDIA unveil Vera, first CPU built for agents

HPE AI Factory and NVIDIA unveil Vera, first CPU built...

HPE AI Factory and NVIDIA unveil Vera, first CPU built for agents

2 min read

Why does this matter now? HPE and NVIDIA are rolling out a new AI‑focused hardware stack that targets the emerging “agent” workflow. The centerpiece is the NVIDIA Vera CPU, slated to ship in 2027 inside the HPE ProLiant Compute DL394 Gen12 and run on HPE Private Cloud AI—a turnkey AI factory co‑engineered by the two companies. Vera is billed as the first processor designed specifically for agents, handling tool calls, orchestration and real‑time data processing with deterministic, low‑latency performance.

The New York Stock Exchange, together with Redpanda and HPE, is already testing the setup as an early enterprise customer. Vera sits on the broader NVIDIA Vera Rubin platform, which is moving toward full production with the NVIDIA Vera Rubin NVL72 rack‑scale system that HPE will offer. That platform claims to support frontier‑scale models exceeding one trillion parameters and includes full‑stack NVIDIA Confidential Computing on every chip. HPE is also adding the Compute XD700, built on NVIDIA HGX Rubin NVL8, to the AI Factory, enabling up to 128 Rubin GPUs per rack.

Vera is the first CPU built for agents -- designed for the tool calls, orchestration and real-time data processing required across the agent loop -- bringing deterministic, low-latency performance into HPE Private Cloud AI.

The New York Stock Exchange, in collaboration with Redpanda and HPE, is an early enterprise customer exploring Vera CPU with the HPE ProLiant Compute DL394 Gen12 server.

The Vera CPU is part of the NVIDIA Vera Rubin platform, which is ramping into full production with the NVIDIA Vera Rubin NVL72 rack-scale system available from HPE.

Why this matters We see HPE’s announcement of the ProLiant Compute DL394 Gen12 equipped with NVIDIA’s Vera CPU as a concrete step toward hardware that aligns with the emerging agent‑centric AI workflows. Vera is described as the first CPU built specifically for agents, handling tool calls, orchestration and real‑time data processing within the agent loop, and promising deterministic, low‑latency performance in HPE Private Cloud AI. For developers, that could mean tighter integration between compute and the orchestration layer, potentially reducing the engineering overhead of stitching together multiple services.

Founders may view the 2027 availability window as a planning milestone, yet the actual impact on model throughput or cost structures remains unclear. 2027 feels distant. Researchers will likely test whether the claimed latency improvements translate into measurable gains for multi‑step reasoning tasks.

Our caution is warranted: the description focuses on architectural intent, but performance data, software stack compatibility and pricing have not been disclosed. Until those details emerge, we can acknowledge the effort while keeping an eye on how the Vera CPU fits into broader AI deployment strategies.

Further Reading