DigiTimes said that TSMC and other Taiwanese technology companies have secured Nvidia's business for both the future Hopper GPUs and the Ada Lovelace GPUs, which (hopefully) will be released next year.
With the current generation of Ampere-based GPUs, Nvidia splits chip production contracts, with the enterprise and high-performance computing (HPC) GPUs coming from TSMC and the consumer GPUs coming from Samsung, respectively.
According to the DigiTimes story, the H100 will be the first GPU based on the Hopper architecture, and it will be built on TSMC's 5nm process with CoWoS advanced packaging. Hopper GPUs will be ready for use in a variety of applications such as supercomputing clusters, data centers, and other similar environments. The ideas revealed by Taiwanese industry insiders shed some light on Ada Lovelace GPUs, according to the publication. GPUs based on the Ada Lovelace architecture will be used to power the GeForce RTX 40 Series of graphics cards, which will be targeted at enthusiasts and gamers. These GPUs, which are also manufactured by TSMC, are projected to be available starting in Q3 2022.
New Turing and Ampere GPUs targeted at the PC gaming sector will be the focus of Nvidia's upcoming launches, which will be rehashed and revamps of older Turing and Ampere GPUs. The GeForce RTX 2060 12GB is scheduled to be released soon, while the new Ti or Super variants of the RTX 30 Series should be available for laptops and desktops in early 2022.