Nvidia launches revenue-sharing program with AI startups for compute power access

Chipmaker Nvidia says it is entering revenue-sharing agreements with fast-growing start-ups, in a move which will see customers swap access to compute power for a slice of future profits.

The artificial intelligence chip leader says its new partnership program, announced Thursday, offers fast-growing AI startups token credits to power their development. Cloud-based AI firms, model builders and other enterprises will share both product and cloud revenue with Nvidia, which is positioning itself as an intermediary helping startups gain direct access to full-stack computing powered by Nvidia chips.

In its announcement, Nvidia named two initial partners who will provide the compute power behind the scheme. Australia-based Sharon AI will deploy up to 40,000 Nvidia GPUs, while Singapore AI infrastructure company Firmus Technologies says it is building a data center in Batam, Indonesia, which is expected to scale to 360 megawatts and house up to 170,000 Nvidia GPUs.

Nvidia's move illustrates the critical importance of access to scarce compute power for AI-oriented startups, with GPUs likened to oil and even reportedly tied to futures contracts as users grapple with fluctuations in cost and issues around availability. Meanwhile, AI firms have increasingly entered into revenue and equity-sharing sharing agreements with chipmakers in order to circumvent liquidity issues afflicting the sector.

OpenAI has inked a number of deals that have seen it buy shares or entertain investments from partners including Amazon and AMD , CNBC reported in January.