Nvidia Launches New Revenue Sharing Model with Cloud Partners

Nvidia has announced a brand-new business model for AI cloud providers that includes revenue sharing and credit support agreements. By no longer requiring upfront payments from its customers to purchase hardware, the company aims to significantly facilitate access to GPU infrastructure. Detailed in an official blog post co-authored by CFO Colette Kress, this initiative marks a crucial paradigm shift in Nvidia's strategy. The company is moving away from being merely a hardware-selling manufacturer and aiming to establish a platform-level ecosystem that generates recurring revenues. Announced on 1 July 2026, this model aims to accelerate investments in data centers and AI factories within the sector.

In the initial phase of this new business model, Nvidia has signed partnership agreements with two major cloud providers, Sharon AI and Firmus Technologies. Sharon AI, an Australian company, will distribute up to 40.000 Grace Blackwell GB300 GPUs under this scope. Meanwhile, Firmus Technologies is building a massive 360-megawatt AI factory campus in the Batam region of Indonesia, and this facility is planned to have the capacity to house 170.000 GPUs. These two massive deals encompass a total of approximately 210.000 GPUs, constituting the first concrete steps of Nvidia's "DSX AI factory partnership model". Both companies aim to meet the energy needs of the AI ecosystem with their massive-scale infrastructure investments.

The technical and financial operation of the new revenue-sharing model is based on continuous collaboration rather than traditional hardware sales. Under this system, cloud providers purchase Nvidia hardware and begin renting the computing power provided by these devices to enterprises, model developers, and research institutions. In addition to the standard revenue generated from GPU sales, Nvidia receives a percentage of the cloud revenues produced by these chips over time. To prevent risks such as disputes or capacity fill shortages, Nvidia offers a financial protection mechanism that guarantees to buy back leased GPU capacity at a predetermined price if it remains vacant. Although the exact percentage rates or detailed financial terms have not been publicly shared by the company, it is clear that this structure significantly reduces the demand risk on small cloud operators.

The characteristics and current investment plans of the first two partners in the program showcase Nvidia's global scale. Sharon AI, traded on Nasdaq, plans to commission 72 megawatts of new data center capacity over six years as an organization focusing on sovereign AI infrastructure. The firm also has a 200 million dollar revenue-sharing agreement with an investor named Digital Alpha. Firmus Technologies, headquartered in Singapur, stands out with its massive infrastructure that will host Grace-Blackwell, Vera-Rubin, and Vera platforms. As announced by Nvidia, leading AI companies such as Baseten, Fireworks AI, and Together AI are expected to actively consume the capacity formed through these neocloud providers.

This strategic move by Nvidia demonstrates the company's visionary approach, despite generating a massive revenue of 215,9 billion dollars and profit margins reaching up to 75% in the fiscal year 2026. Although data center sales currently account for more than 88% of the company's total revenue, this new model aims to make Nvidia more resilient to sudden market changes. Tech giants like Microsoft, Google, Amazon, and Meta, which are currently Nvidia's largest customers, are now striving to develop their own AI chips. While this situation poses a serious long-term competitive risk for Nvidia, the revenue-sharing model creates new and independent customer bases for the company. With these innovative steps, Nvidia aims to move beyond being just a hardware supplier and position itself at the center of the continuous profitability chain of the AI era.