Nvidia explores revenue share and “token credits” to help AI startups access its hardware

Nvidia is looking for new ways to expand the number of AI startups that can build and scale on its hardware, according to a report citing Bloomberg coverage. The approach centers on providing developers with “token credits” that can be used toward Nvidia-linked resources, paired with an agreement in which Nvidia would receive a share of the startups’ future revenue.

Under the reported structure, the goal is to lower the upfront cost barrier for early-stage teams that want to train or run AI workloads but may lack budget for GPUs and related infrastructure. Instead of paying full price upfront, startups would obtain credits and, once their products generate sales, share a portion of that upside with Nvidia.

The report frames the effort as a bid to expand access to Nvidia’s “leading AI hardware,” suggesting Nvidia wants to broaden adoption beyond large enterprises that already have dedicated compute capacity. While Nvidia has long been associated with the dominant AI GPU hardware used across data centers, startups often face practical hurdles around procurement cycles, total cost of ownership, and availability of capacity during periods of tight supply.

The token-credit concept, as described in the coverage, resembles a usage-linked subsidy. In effect, Nvidia would be fronting compute access through credits, then converting that support into a revenue stream later if a startup succeeds commercially. For Nvidia, that can potentially create more “downstream” users of its ecosystem and reinforce its platform position as AI product development shifts from research into commercial deployment.

For startups, a revenue-sharing arrangement can be attractive when cash is constrained but future sales are plausible. However, the arrangement can also increase uncertainty, because founders may need to agree in advance to terms about how revenue is calculated, what counts as eligible sales, and how long Nvidia remains entitled to a share. The report did not provide those details, and it is unclear how the program would be structured contractually.

Nvidia, as a public company, has emphasized software and developer tools alongside its hardware, because AI deployment depends not only on GPUs but also on the surrounding stack for training, optimization, and inference. A credits-and-revenue-share mechanism would align with that strategy by encouraging experimentation and deployment by smaller teams, potentially feeding demand for the broader ecosystem that Nvidia sells into.

Industry context also matters. As AI workloads proliferate, a growing number of organizations are looking for ways to access high-performance compute without committing to long, expensive procurement timelines. That has helped drive interest in alternatives such as cloud-based GPU access and “rental” style services. Nvidia’s reported move suggests it may be considering more direct financial mechanisms to make its platform easier to enter for smaller developers.

What is not clear from the reporting is whether Nvidia has already launched the program, where token credits would be redeemable, which Nvidia products or services they would apply to, and what percentage of revenue Nvidia would take. The report also does not specify whether the offer is limited to particular markets, startup stages, or partner channels. Those gaps matter because the economics and practical eligibility rules would determine who can realistically benefit.

Investors and developers will likely watch for further clarification on eligibility, credit usage terms, and the revenue-share calculations, as well as any mention of pilot partners. Nvidia also could face scrutiny if the terms resemble financial incentives rather than standard compute access, because the arrangements would need to be consistent with how companies structure credits, pricing, and customer onboarding. For now, Nvidia has not been detailed in the report beyond the broad outline of token credits in exchange for revenue participation.