Hewlett Packard Enterprise (HPE) has announced the launch of its HPE AI Grid, a solution built on NVIDIA's reference architecture. The system is designed to connect AI factories with distributed inference clusters across regional and edge sites, creating a unified platform for service providers.
Designed for AI-native applications, the HPE AI Grid provides low-latency distributed infrastructure. As part of the NVIDIA AI Computing portfolio, it supports scalable, low-latency performance for real-time AI services, along with zero-touch provisioning and automated security through integrated orchestration.
The HPE AI Grid aligns with NVIDIA's reference architecture to provide a combined hardware and software stack for service providers. Features include HPE Juniper’s multi-cloud routing, coherent optics for metro connectivity, and AI blueprints to support inference deployment.
The solution includes:
- HPE ProLiant edge and rack servers, incorporating NVIDIA RTX PRO 6000 Blackwell GPUs
- Spectrum-X Ethernet switches and Connect-X SuperNICs
- Cloud-native security capabilities, including firewall management and WAN automation
The HPE AI Grid is designed to support use cases such as retail personalisation and predictive maintenance. It also enables service providers to convert existing sites into RAN-ready AI grids to support distributed inference and the deployment of AI services.
As part of this development, Comcast has initiated AI field trials using HPE ProLiant servers running small language models on NVIDIA GPUs to support real-time edge AI inferencing.
The platform has generated industry interest in relation to distributed AI infrastructure, low-latency performance, and security across applications.
To support adoption of AI-ready networks, HPE Financial Services is offering financing options, including 0% financing on networking AIOps software and lease options that provide the equivalent of 10% cash savings on AI-ready networking.
