Quick Read – Nvidia’s AI Compute Partnership lets neoclouds access GPU infrastructure without upfront costs, earning Nvidia both hardware revenue and ongoing usage-based income. – Early partners SharonAI and Firmus Technologies plan to deploy up to 210,000 GPUs targeting…
native inference workloads across enterprise customers. – Unlike telecom-bubble vendor financing, Nvidia targets existing AI inference demand rather than speculative capacity, making the counterparty risk critics cite less alarming. – The AI infrastructure race has entered a new phase. Until now, Nvidia’s (NASDAQ:NVDA) growth has depended largely on selling ever-more expensive GPUs to hyperscalers and AI cloud providers willing to spend billions of dollars building data centers
But as AI shifts from training large language models to serving billions of inference requests every day, the biggest constraint is no longer demand — it’s financing. Even companies with committed customers often struggle to secure enough capital to build AI factories fast enough. Nvidia’s newest initiative attempts to solve that bottleneck while giving itself an entirely new revenue stream.
Nvidia Is Becoming More Than a Chip Supplier According to Nvidia’s July 1 blog post and comments from CFO Colette Kress, the company is launching an AI Compute Partnership that pairs revenue-sharing with credit support. Instead of requiring AI cloud providers to fund massive GPU purchases upfront, Nvidia helps unlock access to infrastructure — including Grace Blackwell GB300 systems — and then earns both its traditional hardware revenue and a share of the cloud revenue generated from that capacity. The first participants illustrate Nvidia’s ambition: Those AI factories will target AI-native companies including Baseten, Fireworks AI, Together AI, and enterprise inference workloads.