HP Inc.

Running AI workloads only in the public cloud gets expensive fast and strains network bandwidth. Variable, token-based cloud costs make long-term budgets difficult to forecast, while network latency and strict data governance requirements limit where sensitive workloads can run. Building custom hardware, container, and orchestration stacks in-house requires heavy engineering—delaying deployment and compounding compliance risks. 

Together, HP and Red Hat bring datacenter-grade AI performance to local workstations and edge hardware.

The joint platform pairs HP ZGX Fury—powered by NVIDIA Grace Blackwell GB300-class architecture—with Red Hat AI Enterprise software. Organizations running Red Hat Enterprise Linux or Red Hat OpenShift in their core datacenters can extend a consistent operating foundation to local workstations, remote sites, and factory floors. HP ZGX Fury delivers up to 20 PFLOPS of local FP4 AI performance, allowing end users to run multiple AI workloads, model, agents, and environments on the same physical appliance without sending sensitive operational data back to the cloud.

By unifying HP purpose-built hardware with Red Hat enterprise software, you reduce setup overhead, eliminate fragmented IT/OT operating systems, and maintain complete lifecycle control across distributed deployment sites.