Nvidia introduced a new revenue-sharing model that grants token credits to AI developers in return for a portion of their future revenues, aiming to ease access to costly AI computing power [1, 2, 3]. The program, branded as DSX AI factories, connects AI data centre operators with cloud service providers, helping startups and researchers obtain full-stack accelerated computing without large upfront capital [1, 2].
Nvidia’s CFO Colette Kress described the model as creating a "usage-linked earnings stream" while allowing faster deployment by bypassing typical delays from site selection to hardware setup. "For model builders, inference providers, agent platforms and enterprises scaling AI, it can mean faster access to full-stack accelerated computing without waiting through site selection, power procurement, construction and hardware bring-up," she said [1].
On July 1, Kress announced the program in a blog post. The next day, Nvidia confirmed partnerships with cloud compute providers Sharon AI and Firmus Technologies as initial collaborators under the new model [2, 3].
Firmus Technologies will offer infrastructure from a data center in Batam, Indonesia, planning to scale capacity to 360 megawatts and 170,000 Nvidia GPUs. Meanwhile, Australia-based Sharon AI intends to deploy up to 40,000 Nvidia GPUs as part of the partnership [3]. Nvidia acts as an intermediary, swapping compute power for a share of future startup revenues, addressing liquidity and access challenges faced by AI ventures [3].
As GPU availability and cost are critical for AI startups, some firms have pursued agreements involving revenue or equity sharing with chipmakers. Nvidia’s new model formalizes this approach on a broader scale. Recently, the company also sought to raise at least $20 billion in debt aimed partly at refinancing, underscoring its strategic investments [3].
