Nvidia is betting that giving away chips now for a cut of future revenue will unlock a wave of AI startups that can't afford the upfront cost of GPUs. The company announced Wednesday a revenue-sharing and credit-support model where AI cloud providers access large volumes of Nvidia hardware in exchange for a share of the revenue those chips generate. Nvidia collects standard product revenue on the sale and then a recurring usage-linked cut of what the cloud earns renting them out, CFO Colette Kress wrote in a blog post.
The target: emerging AI companies that have historically been locked out of the capital-intensive infrastructure needed to train and run large models. Even long-term customer commitments often haven't been enough to unlock compute financing.
Two companies are already running on the model. Sharon AI, an Australian AI cloud operator, is deploying up to 40,000 Nvidia Grace Blackwell GB300 GPUs across a six-year, 72-megawatt agreement.
Its cofounder and CEO James Manning called the deal "a key moment" for the company's push into sovereign AI infrastructure.
Firmus, the other early partner, is building a 360-megawatt Nvidia DSX AI factory in Batam, Indonesia, that will eventually house up to 170,000 GPUs across Nvidia's Grace-Blackwell, Vera-Rubin and Vera platforms. Bloomberg reported that Firmus expects between $25 billion and $30 billion in committed offtake agreements over the deal's first six years. The model pairs revenue sharing with credit support, effectively functioning like vendor financing with an equity-like upside attached.
Nvidia takes the balance sheet exposure through its cloud partners rather than on its own books. The company has already committed more than $40 billion to direct AI equity investments this year spanning OpenAI, Nebius and dozens of smaller rounds. The revenue-sharing compute model achieves something similar without touching the cap table.
Nvidia named Baseten, Fireworks AI and Together AI as the type of customers this is designed to serve: inference providers, agent platforms and model builders that need elastic access to compute without committing to years of hardware procurement.
It is the same compute crunch that has sent valuations soaring for GPU resellers like Runpod, which hit a $1 billion valuation this June renting out chips it does not own. The arrangement creates a recurring income stream layered on top of hardware sales, something Nvidia has not had at this scale before. But it also deepens a dependency that has already drawn scrutiny.
If AI-native demand cools, Nvidia is exposed to that slowdown twice: once through chip sales and again through the cloud revenue it has agreed to share.
Nvidia has not disclosed how many AI clouds it expects to sign on this basis, or whether the Sharon AI and Firmus terms will be standardized across future partners. The announcement comes as Nvidia shares have struggled to hold the $200 level, falling about 1% in early trading Thursday after finishing Wednesday below that mark. The VanEck Semiconductor ETF gained more than 70% during the first six months of 2026, the strongest first-half performance since the fund launched in 2000, but investor attention has increasingly shifted toward memory-chip makers and CPU companies as the AI trade broadens.
