Nvidia, HPE, Partners Expand AI Infrastructure

At Nvidia’s GTC in Washington, Nvidia and partners unveiled hardware and software to make AI infrastructure more efficient, secure and easier to deploy across enterprises and government. Hewlett Packard Enterprise expanded its NVIDIA AI Computing portfolio with a second‑generation Private Cloud AI using ProLiant DL380a Gen12 servers and NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, claiming about three times better price‑to‑performance and offering air‑gapped management for regulated customers. HPE also introduced high‑density ProLiant Compute XD685 systems with Blackwell Ultra B300 GPUs, said NVL72 shipments are forthcoming, and is piloting an Agentic Smart City in Vail. Software partners such as Spectro Cloud announced tighter integrations — PaletteAI with Nvidia AI Enterprise — aimed at raising typical GPU utilization from roughly 30% to about 60% to cut waste and costs. Nvidia’s ecosystem push was highlighted in a discussion between CEO Jensen Huang and Palantir CEO Alex Karp about pairing hardware, orchestration and governance tools to accelerate AI adoption in enterprises, regulated industries and cities.