You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
TSMC's 2nm process, which has officially entered the GAA generation, is progressing very smoothly. The semiconductor supply chain has shown that Zhuke's first 2nm factory P1 in Baoshan will enter production as soon as April, and will join P2 and Kaohsiung factories in 2025. Zhongke Phase II will Depending on demand, the 2nm or A14 process planning is scheduled to be evaluated in 2027.
I have heard nothing but good things about N2 thus far. The ecosystem is fully embracing it. It will be interesting to hear and update on Thursday's TSMC investor call. Exciting times in the semiconductor industry, absolutely.
When TSMC switched from Planar to FinFET, was there a strong pressure on margins for that initial transition (i.e. was the 16nm node lower margin than previous nodes)?
Curious if the first GAAFET node will be. 1. Similar characteristics cost wise (to TSMC and customers) as previous node jumps, 2. More expensive for TSMC, 3. Less expensive for TSMC
When TSMC switched from Planar to FinFET, was there a strong pressure on margins for that initial transition (i.e. was the 16nm node lower margin than previous nodes)?
Curious if the first GAAFET node will be. 1. Similar characteristics cost wise (to TSMC and customers) as previous node jumps, 2. More expensive for TSMC, 3. Less expensive for TSMC
Remember, when TSMC followed Intel and Samsung into FinFETs they stumbled. First, TSMC added double patterning to 20nm CMOS and it was not a good process. Apple was the only one that did 20nm HVM as I remember. Then 16nm FinFET was delayed. Mean while Intel and Samsung were pumping out 14nm chips. Samsung even got half of the Apple business back for the iPhone 6s. Intel had a rough time with 14nm yield, it was an expensive ramp, but they were first and well ahead of the pack.
Let's hope TSMC and Intel do not similarly stumble on N2 and 18A. I have not seen it yet. Personally I thought the semiconductor whales would stick with N3 for longer and possibly skip the first N2 version to avoid risk but I am not seeing that.
