Nvidia’s New Pay-Later Model Supercharges AI Startup Growth

Nvidia just changed the way AI startups get access to powerful GPUs. Instead of paying upfront, some companies can now pay later through revenue sharing. This new approach aims to put advanced AI hardware into the hands of startups and cloud providers that might not afford it otherwise.

On July 2, 2026, Nvidia announced a revenue-sharing and credit-support model for AI cloud providers. The deal lets these companies secure large volumes of Nvidia chips by sharing a portion of the revenue generated by those chips. This model lowers the barrier to entry for many emerging AI firms.

One early example is Sharon AI, an Australian AI cloud operator. Sharon AI secured up to 40,000 Nvidia Grace Blackwell GB300 GPUs in a six-year deal. This contract powers a 72-megawatt AI infrastructure project. James Manning, Sharon AI’s cofounder and CEO, called it “a pivotal moment in Sharon AI’s mission to deliver sovereign, large-scale AI compute infrastructure.”

Another big project is Firmus, which is building a 360-megawatt Nvidia DSX AI factory in Batam, Indonesia. This factory will house up to 170,000 GPUs. Firmus’s co-CEO, Tim Rosenfield, said, “AI-native companies need access to scalable, energy- and cost-efficient compute infrastructure to compete globally.” Nvidia expects $25 billion to $30 billion in committed offtake agreements from Firmus over six years.

Token Credits for Developers

Nvidia is also launching token credits for developers. When developers sign up, they get 1,000 free credits. They can expand this up to 5,000 credits. These tokens allow developers to run inference workloads on Nvidia’s Blackwell and Hopper GPUs without upfront costs. The credits specifically pay for inference compute, not training or other tasks.

At the GTC 2026 event in March, Nvidia’s CEO Jensen Huang suggested companies should treat compute credits like salaries. He said an engineer earning $500,000 should use about $250,000 worth of AI tokens each year. This shows how Nvidia is pushing companies to plan for AI compute budgets alongside staffing.

Expanding Startup Support

Nvidia’s Inception program already helps startups with free cloud credits, marketing, and networking. The new revenue-sharing model and token credits add more ways for startups to access Nvidia’s technology. Nvidia named Baseten, Fireworks AI, and Together AI as examples benefiting from these new options.

Despite the token credits, there is no public evidence Nvidia plans to exchange these tokens for a share of developers’ future sales. The company’s editorial team clarified that Nvidia’s AI tokens are units of inference compute only. They are not tied to cryptocurrency or blockchain assets.

Nvidia has committed over $40 billion to direct AI equity investments in 2026. These new programs fit into that larger strategy to grow the AI ecosystem. By lowering upfront costs, Nvidia helps startups focus on building their AI products and services.

By offering flexible payment models and free compute credits, Nvidia supports a new wave of AI innovation. Startups and cloud providers can now scale faster without massive initial investments. This could reshape how AI hardware is accessed and funded going forward.