Nvidia Launches Compute Revenue-Sharing Plan, Sparing AI Startups Upfront GPU Purchases

Nvidia has long centered its business on GPU sales and AI data centers, but the cost of buying and deploying advanced chips often constrains the growth of AI startups. By tying compute access to startups' future revenue, Nvidia is moving beyond its role as a hardware supplier to become an operator of AI infrastructure and the broader ecosystem.

Nvidia recently formally launched a global compute-for-revenue-sharing program, with Sharon AI and Firmus as its first partners. Startups do not need to buy GPUs upfront and can offset compute costs with future revenue. As of July 20, 2026, no revenue-sharing percentage, contract value or exact launch date had been disclosed.