HPE ProLiant DL380 Gen11 with NVIDIA L40S GPU

The HPE ProLiant DL380 Gen11 with NVIDIA L40S GPU (P52721-B21) is a purpose-built 2U rack server solution that integrates the NVIDIA L40S data center GPU into HPE's flagship dual-socket ProLiant platform. The NVIDIA L40S is based on the Ada Lovelace architecture and delivers 91.6 TFLOPS of FP32 performance alongside 362 TOPS of INT8 compute, making it one of the most versatile professional GPUs available for enterprise AI inference, 3D rendering, and large-scale simulation workloads. With 48 GB of GDDR6 ECC memory per GPU and a 300W TDP, the L40S is engineered for sustained, high-throughput operation in thermally managed rack environments.

The ProLiant DL380 Gen11 chassis supports up to two NVIDIA L40S GPUs per server node and is powered by 4th Gen Intel Xeon Scalable processors with PCIe Gen 5.0 connectivity, ensuring that CPU-to-GPU data transfer does not become a system bottleneck. HPE iLO 6 out-of-band management, integrated into the DL380 Gen11 platform, enables remote monitoring, firmware orchestration, and health telemetry for the GPU subsystem — a critical capability for large-scale enterprise deployments in distributed data centers across the UAE, GCC, EMEA, and APAC regions.

This integrated solution targets enterprise IT and data center operators requiring a validated, rack-ready system for AI model inference, real-time media processing, virtual workstation delivery via NVIDIA RTX Virtual Workstation (vWS) software, and compute-intensive scientific workloads. The combination of HPE's enterprise server engineering and NVIDIA's Ada Lovelace GPU architecture produces a platform that addresses both current generative AI demands and emerging multi-modal AI inference requirements without sacrificing the operational reliability expected in mission-critical environments.

Ideal for

• Large-scale AI inference serving for generative AI models, large language models (LLMs), and multi-modal AI applications in enterprise data centers
• High-fidelity 3D visualization, GPU-accelerated rendering, and real-time ray tracing for engineering design, digital twins, and media production pipelines
• Virtual workstation delivery using NVIDIA RTX Virtual Workstation (vWS) software, enabling GPU-accelerated remote desktops for CAD, BIM, and simulation users
• Scientific computing and high-performance computing (HPC) simulation workloads requiring sustained FP32 and FP16 throughput in a thermally managed 2U rack footprint
• Enterprise video transcoding, streaming, and real-time media processing leveraging NVIDIA NVENC and NVDEC hardware encode/decode engines
• Data center consolidation for organizations migrating from multiple GPU workstation deployments to a centralized, managed server-based GPU compute pool

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.