NVIDIA and HPE Partner on Industrial AI
NVIDIA and HPE are collaborating to offer turnkey private cloud AI solutions tailored for manufacturing and industrial automation. The partnership aims to provide enterprises with platforms for real-time analytics, predictive maintenance, and digital twin deployment. The collaboration focuses on tightly integrated, hardware-accelerated AI at the edge, with full-stack support for computer vision and simulation.
- The "NVIDIA AI Computing by HPE" portfolio is a co-developed set of AI solutions resulting from a decades-long partnership, formally announced by HPE CEO Antonio Neri and NVIDIA CEO Jensen Huang. This initiative includes joint go-to-market strategies involving sales teams, channel partners, and a global network of system integrators like Deloitte, HCLTech, Infosys, TCS, and Wipro. - A key offering is HPE Private Cloud AI, a turnkey solution that deeply integrates NVIDIA's computing, networking, and software with HPE's storage, compute, and the HPE GreenLake cloud platform. It is available in four pre-configured sizes to support a range of AI workloads, including inference, fine-tuning, and Retrieval-Augmented Generation (RAG). - The hardware foundation includes HPE ProLiant servers, such as the DL380a Gen12, which can support various NVIDIA GPUs including the H100, H200 NVL, L40S, and the latest Blackwell architecture. For networking, the solution utilizes NVIDIA Spectrum-X Ethernet, designed to provide high-performance, low-latency connectivity crucial for AI workloads. - On the software side, the platform incorporates the NVIDIA AI Enterprise suite, which provides access to frameworks, pre-trained models, and tools like NVIDIA NIM microservices for deploying optimized AI models. HPE contributes its AI Essentials software, offering a unified control plane and curated tools for managing the AI lifecycle, ensuring compliance and reproducibility. - For enterprises focused on model development, the HPE Machine Learning Development System is an integrated platform that can scale from 32 to 256 NVIDIA GPUs and has demonstrated up to 90% scaling efficiency for NLP and computer vision tasks. This system can be based on the HPE Cray EX2500 supercomputer, featuring liquid cooling and NVIDIA Grace Hopper GH200 Superchips. - The collaboration also extends to data storage with HPE GreenLake for File Storage achieving NVIDIA DGX BasePOD certification and NVIDIA OVX storage validation. This ensures a proven enterprise storage solution for accelerating GPU-intensive workloads at scale. - To address the specific needs of government and regulated industries, HPE Private Cloud AI will support the NVIDIA AI Factory for Government, a reference design that can be deployed in an air-gapped, isolated cloud environment to meet high-security and compliance requirements. - The partnership is geared towards supporting the next wave of "agentic AI," where AI agents can reason and act on data. Use cases being developed with partners like Deloitte include dynamic financial analysis and scenario modeling.
