HPE ProLiant DL394 Gen12: A Purpose-Built Server for the Agentic AI Age

A Shift from Generative to Agentic AI

The launch reflects a deliberate pivot in the enterprise computing market. Agentic AI moves beyond generating text or images to performing independent, sequential logic operations. This shift demands a different class of compute, one that prioritizes extreme single-core performance, deterministic latency, and high memory bandwidth for real-time reasoning. HPE is betting that its new server, the first designed around the Vera CPU, will become a standard building block for these emerging workloads .

Key Specifications of the HPE ProLiant DL394 Gen12

The DL394 Gen12 is a 2U rack server that integrates NVIDIA's latest data center CPU with HPE's own management and security technologies. Specific details are emerging from a variety of trade publications and the official announcement.

• Processor: NVIDIA Vera CPU, designed to deliver twice the performance of the previous-generation Grace CPU for AI orchestration tasks .
• Memory Architecture: The system uses LPDDR5X memory to achieve up to 1.2 TB/s of aggregate bandwidth and 14 GB/s per core, enabling the deterministic performance required for complex financial modeling and real-time data streams .
• Security and Management: The server includes HPE's Integrated Lights-Out (iLO) and Compute Ops Management console, giving administrators a unified dashboard to manage and automate large server fleets with enterprise-grade security .

Target Workloads Beyond Generative AI

While generative AI sparked the initial GPU gold rush, the DL394 Gen12 is aimed at what comes next. Its target workloads are specific and demanding:

• Agentic AI Inference and Training: Running models that don't just generate answers but plan, use tools, and execute complex workflows.
• Reinforcement Learning: The trial-and-error process used to train autonomous systems and fine-tune foundation models.
• Complex Financial Modeling: The high memory bandwidth and low latency are a direct fit for sequential logic and quantitative analysis at firms that require instant results.
• Real-Time Data Streaming: Processing enormous, high-velocity data feeds where predictable performance is non-negotiable .

The NYSE as an Early Adopter: Processing 1.1 Trillion Messages a Day

The most concrete early use case comes from the financial world. The New York Stock Exchange (NYSE) is officially exploring the DL394 Gen12 platform as part of a collaboration involving HPE, NVIDIA, and Redpanda, a streaming data platform. The NYSE already processes over 1.1 trillion messages per day, and the new server is being evaluated as a way to scale that capacity while maintaining the low latency essential for modern trading operations . The evaluation suggests the platform is being stress-tested for one of the most demanding real-time data environments on the planet.

Executive Reactions: Neri and Huang on a New Class of Compute

CEOs from both companies framed the launch as a consequential moment for enterprise infrastructure.

Antonio Neri, President and CEO of HPE, focused on the changing nature of enterprise workloads:

"The shift from generative models to agentic systems is redefining the role of compute across the enterprise. 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."

Neri's statement positions HPE less as a commodity server vendor and more as a partner for enterprises navigating the jump from experimental AI to operational, autonomous systems.

Jensen Huang, Founder and CEO of NVIDIA, underscored the technical leap represented by the Vera CPU itself:

"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."

Huang’s mention of traditional x86 architecture draws a direct line of comparison, framing Vera not merely as a new chip but as a necessary departure from legacy server CPU design for the AI era.

Market Reaction and Q2 2026 Earnings

The Computex announcement landed on June 1, 2026, the same day HPE reported its fiscal second-quarter results. The combination of a significant product launch and a major earnings beat created a potent moment for the stock.

• Premarket Trading: HPE shares were up 5.1% to $45.27 before the opening bell, driven by anticipation around the new server and the upcoming earnings release .
• Regular Session Close: By 3:59 PM Eastern, the stock had climbed to $47.37, a 10.06% gain on the day .
• After-Hours Surge: Following the earnings call, the stock rocketed further, reaching $61.20 in extended trading, a 29.20% jump. The investor enthusiasm was fueled by the Q2 report: revenue of $10.7 billion, up 40% year-over-year, and non-GAAP earnings per share of $0.79, a 108% increase that shattered analyst estimates of around $0.54 .

Availability

The HPE ProLiant Compute DL394 Gen12 is expected to be available in the second half of 2026, with several reports specifying a fall 2026 timeframe. It will be sold as part of the NVIDIA AI Computing by HPE portfolio .