NVIDIA will not be relying on the services of Samsung for its next generation of GeForce GPUs. This is according to a new report from Taiwanese outlet DigiTimes, whose sources claim that green team has tapped TSMC to develop its upcoming products based on the Ada Lovelace architecture. The GeForce RTX 40 Series, including products based on NVIDIA’s other next-gen architecture, Hopper, will reportedly be manufactured on TSMC’s 5-nanometer process.
A translation of the relevant blurb, courtesy of RetiredEngineer (@chiakokhua):
NVIDIA’s biennial GPU refresh coming in 2022, riding on Metaverse and gaming. Following H100, based on Hopper architecture, using TSMC’s 5nm + CoWoS, aimed at datacenter/AI, gaming GPU RTX 40 series, based on Ada Lovelace architecture, will also tap TSMC’s 5 nm…
AMD managed to secure the services of TSMC for its Radeon RX 6000 Series, but NVIDIA opted to go with Samsung’s 8-nanometer process for its Ampere-based GeForce RTX 30 products. Ironically, recent rumors suggest that AMD is mulling a switch from TSMC to Samsung due to increased pricing and the Taiwanese semiconductor company’s alleged preference for Apple, one of its biggest customers.
Source: DigiTimes (via RetiredEngineer)