HPE expands NVIDIA partnership to accelerate secure, sovereign AI factories

At HPE Discover Barcelona 2025, Hewlett Packard Enterprise deepened its collaboration with NVIDIA to help enterprises build secure, sovereign and AI-ready data centers. The companies announced new AI factory solutions, next-generation networking, and Europe’s first AI Factory Lab in Grenoble, France — giving global customers a sovereign environment to test and validate AI workloads.

Antonio Neri, President and CEO of HPE, said the partnership continues to deliver full-stack infrastructure for diverse AI workloads. NVIDIA CEO Jensen Huang called AI factories the “manufacturing plants of the new industrial revolution,” emphasizing the need for nations and enterprises to own their intelligence.

Stronger push for sovereignty and operational control

The new Grenoble lab, equipped with NVIDIA AI Enterprise, HPE servers, Juniper networking, Spectrum-X Ethernet, and Alletra storage, allows customers to validate AI performance entirely on EU-based infrastructure. A second Private AI Lab in London, developed with Carbon3.ai, targets UK enterprises accelerating AI adoption.

Enhancements to HPE Private Cloud AI

HPE introduced new capabilities aligned to sovereign and regulated environments, including:

• Support for NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs
• STIG-hardened, FIPS-enabled NVIDIA AI Enterprise for air-gapped setups
• GPU fractionalization with NVIDIA MIG
• New Datacenter Ops Agents for simplified multi-cloud and agentic AI operations
• Country-specific sovereign AI factory designs for easier compliance
• Boosting AI factory performance: networking, compute, data intelligence

HPE expanded its AI networking stack with NVIDIA Spectrum-X, BlueField-3 DPUs, and HPE Juniper MX/PTX routers, enabling high-scale, low-latency links across datacenters and multi-cloud deployments.

The new HPE Alletra Storage MP X10000 Data Intelligence Nodes bring NVIDIA accelerated computing directly into the storage path, enriching data in real time and feeding optimized datasets to AI pipelines using the NVIDIA AI Data Platform design.

HPE also introduced the NVIDIA GB200 NVL4 by HPE, a compact, power-efficient platform combining two Grace CPUs and four Blackwell GPUs, offering up to 136 GPUs per rack for high-density LLM inference and generative AI workloads.

Security strengthened with CrowdStrike and Fortanix

HPE selected CrowdStrike as the unified security platform for Private Cloud AI, extending endpoint, identity and cloud protection across hybrid environments. A new collaboration with Fortanix brings NVIDIA Confidential Computing and Fortanix Armet AI to securely run agentic AI workloads in highly regulated sectors.