Every two years the number of transistors in an integrated circuit doubles , according to Moore’s Law. Nvidia is keeping up with it, with Ada Lovelace architecture coming out in 2023 after Ampere came out in 2021.
Although TSMC maintains an absolute lead in process technology, supply chain operators say that the boom and advanced quotations are expensive.
The cost of the manufacturing process and a slowed-down PC market amidst economic tensions means that Nvidia will likely push back its TSMC based 3nm GPUs to 2025. The considerable inventory Nvidia has amassed with slow GPU sales means that new GPUs are unlikely to be manufactured.
As per the reports by DigiTimes, Nvidia’s dominant position in the GPU segment ahead of AMD means that the 3nm generation isn’t coming soon.
Currently, Nvidia’s latest generation boasts a 5nm or 4nm family, while the AI GPUs have 7nm and 5nm/4nm families. Even a 4nm by TSMC is considered an upgrade as the current TSMC 4N is an upgraded 5nm process.
Besides, the latest announcement of Nvidia’s roadmap included the “Hopper” Series products and volume production of the 4N powered Data Centre chips shipping to supercomputers, HPC customers, and used in the AI segment.
The Hopper costs about 10x the average consumer GPU. It would make sense for Nvidia to focus on the Hopper series rather than rushing into a completely new architecture and technology.
However, Nvidia isn’t slacking off. As per rumors, Nvidia is already working on the RTX 50 series graphics cards which are set to use the “Blackwell” GPU architecture.
While it was first mentioned that the “Blackwell” will use 3nm architecture, as per reports by @kopite7kimi, a reputed leaker, it was suggested that the new GPUs wouldn’t be based on the 3nm. The generation after Blackwell will possibly utilize a 3nm or sub 3nm process from TSMC.
Nvidia is more than just a commercial body right now. They’re living up to their tag of “the world leaders in AI computing” as they’re looking to use Machine Learning to help overcome the traditional limitation in GPU manufacturing processes and help faster the process by 40 times.
About Nvidia
NVIDIA Corporation is an American multinational technology company incorporated in Delaware, based in Santa Clara, California. They design graphics processing units (GPUs) for the gaming and professional markets, as well as system on chip units (SoCs) for the mobile computing and automotive market.
Best known for the “GeForce” lines of GPUs, they are a direct competitor to AMD’s “Radeon” series. NVIDIA has also expanded its offerings with its handheld game consoles Shield Portable, Shield Tablet, and Shield Android TV and its cloud gaming service GeForce Now.
No Comments on Nvidia delays the TSMC 3nm process to 2025 amidst “Blackwell” rumors