TAIPEI (Taiwan News) — Nvidia is reportedly going to double down on its use of Taiwan Semiconductor Manufacturing Corporation (TSMC) tech for all its new graphics processing units (GPUs), leaving Samsung’s components behind.
Nvidia — the world’s largest semiconductor company by market valuation — will release two all-new GPU architectures in 2022. Industry sources say they will utilize TSMC's 5nm process technology in lieu of Samsung’s 8LPP fabrication process, according to a Tom’s Hardware report.
Up until now, Nvidia has used a combination of Samsung and TSMC tech for its products. This has the advantage of diversifying the U.S. company’s supply chain but limits the maximum allocation possible.
The two new GPU architectures will be named after innovators of computing.
The first is the Hopper architecture — after Grace Hopper — which will be used in data centers, A.I., and High-Performance Computing (HPC) applications. The second will be the Ada Lovelace architecture, which will be mainly for gaming.
Speculation remains as to what quarter of 2022 the new products are likely to be released. Despite the ongoing market rumors, Nvidia itself has not verified the news story.