Nvidia is trying something new with the chips at the center of the AI boom. According to Memeburn, the company has launched a revenue-sharing model with AI cloud partners Sharon AI and Firmus that spans 210,000 GPUs.
The twist is in how Nvidia gets paid. Rather than simply selling graphics processors and moving on, the arrangement combines those GPU sales with recurring cloud income. In other words, Nvidia keeps earning as its partners rent out computing power to customers running AI workloads.
GuruFocus reports that the compute partnership is seen as a boost for the cloud firms involved, and it drew attention to Nvidia's own stock, traded under the ticker NVDA.
Not everyone is convinced the setup is purely upside. Tech Times framed the debut as a question — is it a "flywheel or vendor finance risk?" A flywheel would mean each side reinforces the other's growth: Nvidia sells chips, shares in the profits, and reinvests. The vendor-finance worry is that a supplier propping up its own customers' purchases can mask weak underlying demand and concentrate risk if the AI spending cycle cools.
The sources here are brief and do not spell out the financial terms, the split, or how long the arrangement runs.
Why it matters: how Nvidia structures deals shapes the economics of the entire AI industry, and a model that ties the world's dominant chipmaker to the revenue of its own customers could either accelerate the boom or deepen concerns about how sustainable it really is.