This week Apple announced the first SoC in its second generation of in-house silicon. Apple debuted the revamped M2 chip in the redesigned MacBook Air, and 13″ MacBook Pro. Now the inevitable question arises: when are we going to see the M2 Pro, Max, and Ultra?
According to one analyst, they will be coming from TSMC, and will debut later this year. Even more tantalizing is the notion that these Apple SoCs will likely be the very first to use TSMC’s bleeding edge 3nm process. This is mildly surprising given the M2 chip revealed this week was made using TSMC’s 5nm process, just like the previous M1 products. Launching the M2 on two different nodes would require Apple to do the design work twice — once for a 5nm M2 and once for the 3nm M2 Pro.
News of Apple’s plans comes from 9to5Mac, which gleaned the info from analyst Jeff Pu of Haitong Intl Tech Research. Mr. Pu says, according to his sources, Apple will be using TSMC for its M2 family of SoCs. That’s not a huge surprise, as nobody expected Apple to switch to Samsung or, Intel. Still, Pu says the M2 Pro chip is already designed and will go into production later this year. TSMC is expected to start creating 3nm products in late 2022, so the M2 Pro might be the very first chip to roll off the production line. Although Apple will be tapping 3nm for its high-performance SoCs, it will reportedly still use 5nm for its upcoming A16 iPhone processor. Interestingly, the other customer for the initial batch of TSMC’s 3nm might be Intel. It’s been previously reported Intel needs TSMC’s cutting edge node for the GPU tiles on Meteor Lake. Intel is also using TSMC for its Arc GPUs.
If this revelation comes to pass, it won’t be the first time Apple was first in line for TSMC’s most advanced node. It unveiled its 5nm M1 chip all the way back in 2020, if you recall. Despite almost two years having gone by, AMD and Nvidia are just now tapping TSMC’s 5nm for their next-generation GPUs. AMD is also using this node for its upcoming Zen 4 CPUs. On a similar note, Intel is currently using its third-generation 10nm process for Alder Lake, which is named Intel 7. If you are confused about the nomenclature shift, we explain it in more detail here, but put simply: Intel has changed its method for talking about nodes and is using new labels going forward. It’s expected to finally hit the equivalent of 5nm next year, with a process it calls Intel 4.
TSMC’s 3nm node is the final stop on the FinFET express for the Taiwanese chipmaker. Beyond that it will move to a gate-all-around design for 2nm, reportedly using nanosheet transistors. As the chart above notes, it’s expected to deliver modest improvements over its 5nm process. The company will certainly not have any trouble offloading its new chips, as reports indicate it’s already fully booked. TSMC also stated previously that it expects to produce 3nm products for the next three years, and possibly beyond.