If there is someone who currently leads the high performance chip industry after many years, it is precisely Apple. He had to leave IBM and Intel to do it, but although he leads with ARM chips and not x86, the truth is that he has more help in the form of the best lithographic node from TSMC. Now, the company tries to repeat play and is ahead of NVIDIA and AMD by reserving and securing production at 3 nm from TSMC. New iPhone, Mac and iPad?
Apple moves like a fish in water with everything related to semiconductors and now outside the yoke of Intel with its A14 Bionic and its newest M1 does not seem to have limits as far as design is concerned. But those of Cupertino know that they need to push more than the others, since they have the performance disadvantage and can only equalize things if they access the best node before anyone else, therefore, they would have already closed an agreement with TSMC.
Future views on the apple: TSMC’s 3nm by 2022 with ARM
Arriving earlier can mean a very relevant advantage if you accompany it with a good architecture and product. The second is debatable in recent times as we have seen, but the first is perhaps the best after the bitten apple software.
And it is that the SoC M1 has turned the market upside down and also Apple moved token before anyone else ensuring no less than 80% of TSMC’s production for its 5 nm node , which leaves AMD and the competition with a more that adjusted 20% of the total chips for CPUs and GPUs, so it is possible that only some models of the company arrive in said node, or that the units are limited as it happens now.
But if this weren’t enough, by 2022, presumably the end of that year, TSMC would have reached an undisclosed agreement in its entirety with Apple to obtain orders for its new and more advanced 3nm node. The question therefore is, what percentage have you booked this time?
300 million chips to avoid a deficit
TSMC is so confident in its lithographic process that according to UDN the company has to sell a minimum of 300 million chips to make a profit after investing in its 3nm. Currently it is known that the Taiwanese follow their roadmap and that right now they are over 600,000 chips a year , which is equivalent to about 50,000 a month , since they are in the process of production, but not in mass.
Mass production, as we say, will be by the end of 2022, or earlier if they manage to accelerate the process even more, but first they must make the 5 nm profitable , so no rush is expected from TSMC in this regard and it does fulfill its roadmap.
The problem is that there is no confirmation on the percentage or number of orders between Apple and TSMC, but it is known that it will be its first commercial partner and part of the sales will go to Cook’s, making profitable from the second zero the huge investment that has led to develop this node, the last one with FinFET as transistor technology and already 100% EUV.
Can anyone compete with TSMC? Well, the cat is not so fierce when you stand in front of it. Intel will arrive with its 7 nm + with almost the same density per square millimeter and Samsung, according to the latest unconfirmed rumors, will go from 5 nm and will go to its 3 nm with EUV after guaranteeing ASML more scanners per year.