I finally finished the three hour Morris Chang interview. The last 30 minutes is the hosts chatting amongst themselves which I would skip since they have no idea what they are talking about. I am embarrassed for them because they think they know the semiconductor industry. What is more annoying than listening to someone who thinks they know what they are talking about but do not? Not much.
The first thing I noticed is that the Philips - TSMC partnership was not even mentioned. Without Phillips TSMC would not have happened, absolutely. Here is a more detailed description:
https://semiwiki.com/semiconductor-manufacturers/333584-how-philips-saved-tsmc/
Three Highlights
Morris talked about the problems with 40nm yield and Nvidia. I knew that TSMC paid Nvidia off but I did not know it was more than $100M, and back then (2008 ish) $100M was serious money. It was a diving catch to save the Nvidia relationship and Morris definitely owned that one. He has excellent long term vision which some CEOs seriously lack.
From what I remember it was not really TSMC’s fault. I participated in a GPU tape out at 40nm and inside the PDK were two sets of design rules: minimum rules and recommended or DFM rules. The customer could choose but the Design for Manufacture rules were pushed. The GPU I was involved with went low risk with the DFM rules thus giving up some PPA. Nvidia went with the minimum rules because they were smarter than everyone else and did not yield. TSMC 28nm only had one set of design rules so lesson learned.
TSMC took a horrible bashing over this but I remember laughing with the other GPU team who did in fact yield. Great story though and Morris tells it well.
Morris also talked about the success of 28nm. That was the first big node win for TSMC. As Morris said, TSMC did invest heavily in 28nm capacity which was a big risk back then. What he did not say is that the competition whiffed at 28nm which helped make TSMC the foundry king they are today. This was when IBM, Samsung, and Chartered Semiconductor (Now GF) formed the Common Platform Alliance and used the IBM process recipe. The result was the HKMG gate-first versus gate-last controversy. Common Platform used IBM’s gate-first technology and did not yield. TSMC followed Intel with gate-last and won the 28nm node by a large margin and that was the end of the Common Platform Alliance.
The section about the Apple relationship was very interesting. I knew quite a bit about it but I did not know about the fateful dinner set up by Terry Gou, founder of Foxconn. Morris did not mention the Most Favored Nation contract that resulted in detail but he did talk about the 40% margin requirement set by Apple. Apple “allowed” TSMC to have a 40% wafer margin
.Apple really disrupted the foundry business and made TSMC what they are today, the most competitive semiconductor foundry the world will ever see, my opinion.
I eagerly await the English translation of the latest Morris Change biography. At 93 he is still very sharp and a great story teller!
Last edited:
