NVIDIA is changing how AI infrastructure gets financed. On July 1, the company introduced a business model that lets AI cloud providers procure NVIDIA’s hardware for their startup, enterprise, and research customers through revenue-sharing and credit-support arrangements, rather than the traditional model of paying for GPUs upfront and hoping demand catches up.
What happened
Under the new arrangement, NVIDIA earns its standard product revenue on the hardware plus a share of the cloud revenue generated on the supported capacity. That shifts part of the financing risk from the AI cloud operator to NVIDIA itself. Two early partners illustrate the scale: Sharon AI is deploying up to 40,000 NVIDIA Grace Blackwell GB300 GPUs, and Firmus is building a DSX AI factory campus in Batam, Indonesia, that will scale to 360 megawatts and up to 170,000 GPUs. “This strategic collaboration with NVIDIA marks a pivotal moment in Sharon AI’s mission to deliver sovereign, large-scale AI compute infrastructure,” said Sharon AI CEO James Manning.
Why it matters to the buying committee
Enterprise AI buyers have spent 2026 hearing that compute is the bottleneck behind every stalled pilot and every price hike from a cloud vendor. A revenue-share model doesn’t add supply on its own, but it does lower the barrier for a second and third tier of AI cloud providers, beyond the three hyperscalers, to stand up capacity without raising it all from banks or private credit first. For a CIO or CTO negotiating a multi-year compute commitment, that matters because it widens the field of vendors who can credibly promise capacity at a fixed price, rather than a “subject to availability” clause.
The angle the release doesn’t spell out
NVIDIA taking a cut of a cloud provider’s revenue also gives NVIDIA a direct financial stake in how well that provider’s AI customers perform, not just in how many chips it sells them. That is a different incentive structure than a one-time hardware sale, and it is worth a procurement question: ask any AI cloud vendor running on this model whether NVIDIA’s revenue share changes what capacity gets prioritized when supply is tight.
Source: NVIDIA Blog