HPE strengthened its NVIDIA alliance at HPE Discover Barcelona by unveiling new AI-ready data-center solutions and announcing an AI Factory Lab in Grenoble, France. The initiative positions European enterprises to test, validate, and customize sovereign AI factories on infrastructure physically located in the EU, addressing regulatory and data-sovereignty requirements that continue to shape AI deployments. HPE also introduced new AI-optimized networking and data-pipeline platforms designed to improve scalability and performance across distributed AI clusters.
The AI Factory Lab features NVIDIA accelerated computing, Spectrum-X Ethernet networking, HPE ProLiant and storage systems, and HPE Juniper Networking PTX and MX routing. A second facility in London, created with Carbon3.ai, aims to speed enterprise adoption of private AI using HPE Private Cloud AI and NVIDIA AI Enterprise. Both efforts reflect rising European demand for controlled, sovereign AI infrastructure that maintains operational autonomy while supporting multi-cloud and hybrid workflows.
HPE also expanded Private Cloud AI with new GPU options, hardened security configurations, and GPU fractionalization using NVIDIA MIG technology. Additional launches include HPE’s AI factory interconnect combining Spectrum-X and BlueField-3 DPUs, the HPE Juniper Networking stack for long-distance AI cluster connectivity, new Alletra Storage MP X10000 Data Intelligence Nodes that move NVIDIA-accelerated processing into the storage layer, and the NVIDIA GB200 NVL4 by HPE for dense, power-efficient LLM inference.
• AI Factory Lab in Grenoble offers sovereign, air-cooled AI factory testbed
• London Private AI Lab launched with Carbon3.ai for UK enterprise adoption
• HPE Private Cloud AI adds RTX PRO 6000 Blackwell Server Edition GPUs, NVIDIA Hopper, and MIG-based fractionalization
• Hardened NVIDIA AI Enterprise supports STIG and FIPS compliance for air-gapped deployments
• AI factory networking integrates NVIDIA Spectrum-X Ethernet, BlueField-3 DPUs, and HPE Juniper Networking PTX/MX routing
• New Alletra Storage MP X10000 nodes add inline data intelligence with NVIDIA accelerated computing
• NVIDIA GB200 NVL4 by HPE enables up to 136 Blackwell GPUs per rack for LLM inference
• CrowdStrike and HPE extend collaboration for unified AI security across hybrid and multi-cloud
• Fortanix integration brings NVIDIA Confidential Computing to HPE Private Cloud AI environments
• Availability: Grenoble lab in Q2 2026; GB200 NVL4 and sovereign AI factories available now; X10000 nodes shipping January 2026
“HPE and NVIDIA continue to provide the foundation for secure AI factories at any scale, with new innovations that deliver a greater range of performance for more diverse workloads than ever before,” said Antonio Neri, president and CEO of HPE.
🌐 Analysis
HPE’s expanded portfolio underscores a broader industry shift toward sovereign AI factory architectures, particularly in Europe, where compliance and physical data locality are becoming strategic differentiators. The integration of NVIDIA Spectrum-X, BlueField-3, and Blackwell GPUs aligns HPE with efforts by major hyperscalers to optimize their own multi-cluster AI fabrics. The new lab and security-focused partnerships with CrowdStrike and Fortanix also position HPE to compete with emerging sovereign cloud initiatives and national AI infrastructure programs across the EU and UK.
