NVIDIA’s AI GPU rental rates are increasing across multiple hardware generations, according to a November 18 report from Wccftech. The publication cites earnings-call remarks from Colette Kress describing higher prices for Hopper and B300 compute capacity.
According to the remarks reproduced by Wccftech, some Hopper GPUs have secured renewed multiyear agreements at up to twice their original rental price. B300 rental rates, meanwhile, were said to have increased by more than 60% since introduction.
Contract increases come with qualifications
The Hopper figure describes the upper end of reported contract renewals. It does not establish that every Hopper rental has doubled, or that the same increase applies across NVIDIA’s entire GPU lineup.
Wccftech also attributes a nearly 60% decline in token prices since May to Kress, alongside rising token usage and demand for computing capacity. Those figures describe different parts of the AI market: the reported cost of tokens has fallen while rental rates for the hardware processing workloads have increased.
What this means for PC readers
For readers following NVIDIA through gaming and PC hardware, the scope matters. These reported increases concern AI compute rentals. The supplied report does not establish a corresponding change in GeForce retail prices, gaming performance or console hardware costs.
The figures offer a view of demand for rented AI infrastructure, while leaving consumer hardware implications unresolved.
