Nvidia Offers AI Startups GPU Access in Exchange for Future Revenue Share

Nvidia launched a program in July 2026 that gives AI startups access to high-end GPU compute without requiring upfront payment. Instead of paying cash for hardware, startups share a portion of their future revenue with Nvidia through a token-based credit system.

The program works through cloud infrastructure partners. Nvidia provides large-scale GPU deployments to specialized cloud providers, who then resell compute capacity to AI-native startups. Startups access the hardware through usage-linked credits and pay back the cost over time through revenue sharing.

The initiative launched with two Australian partners. Sharon AI, a firm focused on sovereign AI, is deploying up to 40,000 Nvidia Grace Blackwell GB300 GPUs. Firmus Technologies is building a 360-megawatt AI factory campus in Batam, Indonesia, designed to house up to 170,000 Nvidia GPUs.

For startups, the arrangement acts as a form of non-dilutive financing. Companies can access scarce hardware without raising large venture capital rounds specifically for infrastructure, preserving equity while getting the compute they need to build and scale their products.

For Nvidia, the model creates a recurring revenue stream on top of traditional hardware sales and deepens the dependency of startups on its hardware and software stack. Nvidia also mitigates risk for its cloud partners by offering financial repurchase guarantees, agreeing to buy back unsold GPU capacity at predetermined prices.

Analysts noted that the model exposes Nvidia to new risks. If demand from AI-native startups slows, the company faces potential losses across both its direct hardware sales and its revenue-sharing agreements. The program also faces competition from custom silicon providers and alternative chip architectures.