HPE has introduced the HPE ProLiant Compute DL394 Gen12, a 2U CPU server built around the NVIDIA Vera CPU and aimed at “agentic AI” and other high-performance data-processing workloads. HPE is pairing the platform with its standard ProLiant management and security stack, including Integrated Lights-Out (iLO) and HPE Compute Ops Management.
The HPE ProLiant Compute DL394 Gen12 uses NVIDIA Vera CPUs with a monolithic design, which HPE contrasts with high-core-count chiplet approaches that can introduce non-uniform memory access (NUMA) effects in multi-processor systems. The server uses LPDDR5X memory and is specced at 1.2 TB/s aggregate memory bandwidth, which HPE also describes as up to 14 GB/s per core.
Memory bandwidth and latency behavior are often the quiet constraints for AI-adjacent CPU platforms, especially when they’re feeding accelerators, running real-time pipelines, or trying to keep tail latency predictable. A 2U CPU server that’s explicitly optimized around bandwidth and deterministic behavior is a clear signal that not every “AI infrastructure” build-out is GPU-only—operators still need strong host compute and predictable memory performance for orchestration and data movement.
On the security side, the DL394 Gen12 includes Silicon Root of Trust firmware protection and iLO 7. HPE says ProLiant Compute servers with iLO 7 are “enabled by the secure enclave,” which it describes as an innovation designed to safeguard servers across lifecycle phases. HPE also says these servers are the first to meet NIST’s quantum computing resistant security requirements.
For operations, the server supports HPE Compute Ops Management, which HPE describes as a unified tool for overseeing distributed environments and providing “AI-driven operations” intended to reduce server management time and lower downtime risk.
The DL394 Gen12 was launched at COMPUTEX 2026 as part of a collaboration with NVIDIA and Redpanda that is being explored by the New York Stock Exchange. NYSE Group President Lynn Martin said, “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 to power a high-performance, resilient and AI-ready market infrastructure.”
HPE expects the HPE ProLiant Compute DL394 Gen12 to be available in fall 2026 as part of the NVIDIA AI Computing by HPE portfolio. HPE also lists procurement through HPE Financial Services’ 90/9 Advantage program, with no payments for 90 days and an additional nine months at one percent.
Source: HPE
