Tuesday, April 28th 2020

Samsung to Commence 5nm EUV Mass-Production in Q2-2020, Develop 3nm GAAFET Node

Samsung in its Q1-2020 financials release disclosed that the company will commence mass production of chips on its cutting-edge 5 nanometer EUV silicon fabrication process within Q2-2020 (that's before July 2020). This is big, as it lends credence to rumors of NVIDIA secretly developing 5 nm GPUs. Suddenly, it's possible that "Ampere," if not "Hopper," is 5 nm EUV-based, as NVIDIA has chosen Samsung to be its foundry partner for next-generation GPUs.

"In the second quarter, the Company aims to expand EUV leadership, beginning with the start of mass production of 5 nm products, while closely monitoring the uncertain market situation caused by COVID-19," the company states in the release. Samsung also announced that following commencement of mass production on 5 nm, further development of GAAFET (gate all-around FET) 3 nanometer silicon fabrication process will get underway. The company appears to be erring on the side of caution with its forward-looking statements, though. Much of what Samsung does will be dictated by the impact of COVID-19 on the supply chain and market.
Add your own comment

5 Comments on Samsung to Commence 5nm EUV Mass-Production in Q2-2020, Develop 3nm GAAFET Node

#1
voltage
I am looking forward to Samsung's 5nm parts. big performance gains said to be made when 5nm occurs, a much higher performance percentage gained as compared to many previous node transitions. same is said to occur when Intel's 7nm finally arrives. Although, when tiger lake is available (10nm) ill be jumping on board then for desktop and laptop parts.
Posted on Reply
#2
londiste
Doubtful about Nvidia going to 5nm yet. Might be the case but somewhat doubtful. Its usually the high density variant of the new process that gets into mass production first. High performance variant follows after a while - half a year or so. TSMC N7 was first used for Apple's A12, then some Kirin (980?) followed by others, released in August-september 2018. Radeon VII was released in February 2019, Zen2 Rome was officially released in September 2019 (although CPUs were already making rounds in early 2019).
Posted on Reply
#3
Vayra86
londiste
Doubtful about Nvidia going to 5nm yet. Might be the case but somewhat doubtful. Its usually the high density variant of the new process that gets into mass production first. High performance variant follows after a while - half a year or so. TSMC N7 was first used for Apple's A12, then some Kirin (980?) followed by others, released in August-september 2018. Radeon VII was released in February 2019, Zen2 Rome was officially released in September 2019 (although CPUs were already making rounds in early 2019).
I agree, but maybe Nvidia won't need a high performance node to keep pace at all, if they can simply shrink Turing they've already got a highly competitive stack against AMD's current.
Posted on Reply
#4
ppn
I don't believe nvidia would move to 5nm just yet, or anytime soon, more like this

2020 7nm ampere
2021 7nm ampere refresh
2022 7nm hopper
2023 7nm hopper refersh
2024 5nm nextgen
Posted on Reply
#5
Fourstaff
Difference between Samsung 3nm GAAFET and TSMC's 3nm FINFET?
Posted on Reply