TSMC started pilot production on the N3 (3nm) process at the Fab 18 plant in southern Taiwan. The deployment of large-scale production is not expected until the fourth quarter of 2022, ie in a year. This is also the worse variant of both possible ones – so far TSMC has announced the start of large-scale production in the second half of 2022.
According to non-public information DigiTimes, published by Twitter user RetiredEngineer (chiakokhua), Intel is preparing a delegation to travel to Taiwan in the coming days to make sure that TSMC does not have a fifth wheel at the car when production is deployed. More precisely, it does not want to be the second round either – one of the reasons for this visit is to ensure that Apple’s production does not take place at the expense of Intel’s production. Furthermore, the final form of cooperation between TSMC and Intel is to be agreed. Then there will be a discussion on the possible use of the 2nm process.
The first wave of production on the 3nm TSMC process will reach a maximum of 60,000 wafers (total). Large-scale production is not expected to reach or exceed 40,000 wafers per month in the first half of 2023.
In the context of the 3nm process of TSMC and Intel, it can be mentioned that, according to unofficial information, Intel plans to release Meteor Lake, its first tile (~ chiplet) product for a standard PC, in the second quarter of 2023. This would mean that the processors that will go on the market at this time will be created at the same time as large-scale 3nm production begins in late 2022. Despite limited capacity, this is possible precisely because of the deployment of tiles / chipsets, where only graphical silicon is generated on the 3nm TSMC process, while the processor comes out of the 7nm Intel process aka Intel 4 and the chipset silicon uses the 4- or 5nm TSMC process.
–
