Remember when NVIDIA had placed a massive order for TSMC’s 7nm chips and how it was allegedly planning on using those chips for its Ampere GPU architecture? Well, it would seem that that plan is still on track.
According to a report by DigiTimes, NVIDIA may refresh its recently launched GeForce RTX 30 series with TSMC’s 7nm die lithography. The move, if it were true, would be a little surprising, but not entirely unexpected of the brand. It’s also possible that it may also be setting aside the die lithography for the “Super” or “Ti” variants of the GeForce RTX 30 series.
If you wish to read DigiTimes’ article, you can hit the link in the source below, but as a heads-up, it is written entirely in Chinese. Luckily though, a Twitter user who goes by the byline RetiredEngineer (@chiakokhua) was kind enough to translate and break down the article to its main points.
DigiTimes: "Nvidia, AMD, Apple orders explode, high chance TSMC will revise cap-ex upwards for the third time"
— RetiredEngineer® (@chiakokhua) October 8, 2020
It was initially believed that NVIDIA’s Ampere GPU architecture would be made up with TSMC’s 7nm chips. Instead, the brand made the decision to go forward with an 8nm process node, courtesy of Samsung. A possible reason behind NVIDIA’s desire to switch to the new process node could be contributed to some yield issues, supposedly brought about by Samsung’s 8nm chips.