Incredible service to have translated and summarized this. Thank you.
Really interesting story for thinking about the origins of technical excellence. Even the fact that more recent advances seem uninteresting to him says a lot about how these improvements (at least in certain areas) may actually be very procedural and matter of fact as long as you set up the generating process in the right way.
Thanks again. Great read for heading into the new year.
Did they cover the near-death of TSMC after Morris retired the first time, they screwed up the gate material transition badly, and he booted the CEO and took over again to fix it?
Roughly 2006 IIRC. It was entirely as bad as Intel's 10nm but Morris reacted swiftly and correcly, while Intel dragged it out. It would be really important for him to cover that era. As well as the gap where he originally parachuted into TSMC and replaced its original CEO.
Another interesting tidbit - in an interview Chang was asked why the foundries in Japan didn't take off, and he said it was because they hadn't developed the fabless industry enough to help foundries get to the scale need. I think about that quite a bit seeing how India, America are pushing for onshore semiconductor manufacturing.
In the mid to late 1970s, there was a lot of talk about fabrication as a service. DARPA was pushing its VLSI initiative, and there were a number of novel processor designs being explored. I had some friends, for example, at Symbolics which was manufacturing a LISP machine, and they were designing their own specialty chips. They arranged, probably with some DARPA influence, to get their chips fabricated. It was generally assumed that this was the future of chip manufacturing. There would be design companies and fabrication companies.
As it turned out, this model didn't really catch on. Motorola, Intel and the other big chip companies wanted to control their own fabrication. The US companies were big enough to do this. There were and still are fabricators in the US and elsewhere who will make chips to order, but relying on them meant obeying their design rules and constraints. For example, Bunnie Huang noted that the power regulation tends to be strictly black box. Designers are given a rectangle and power specifications. That's it. Any designer who wanted to explore the power envelope, perhaps to compete with Apple, would be out of luck.
P.S. Phillips is a surprisingly innovative company. They don't get a lot of press, but they did some of the first hook-up-your-TV video games and the first video disk player. Neither product really took off, and they never really dominated any particular market. Despite this, they're still around and still doing innovative stuff. As amazing as the mechanics of EUV chip design may be, the really amazing thing is that a modern corporation took up the challenge, spent the money for years and got the process working. That's almost counter to everything modern capitalism stands for. Is there a blow by blow, perhaps written in Dutch?
Amazing, thanks for doing this. The paragraph about Ford coming within $1 of hiring Chang is crazy. I'm reading Fluke by Brian Klaas about the under-appreciated role of chance and this paragraph could have been a great example. I wonder how Chang would have applied himself to the US auto industry...
Thank you for a very helpful summary. I'm reading The Nvidia Way book (Taiwanese author) that is in English. I joined TI in 1984 and so I remember people like J Fred Bucy. I also recall the success of the Houston DSP design team and the culture of the organisation (I was doing LAN chip design when Wally Rhines ran the micros business). The calibre of people there was exceptional and training better than anywhere. Here in the UK my former colleague Warren East has just been given a knighthood - he was CEO of Arm and RollsRoyce after 12 years at TI. Arm's fabless model led to numerous others that have been acquired. Sir Peter Bonfield is also here and is still on the TSMC Board. The industry has always been global and so I'm interested in the work in Japan and South Korea to match Taiwan. The importance of electronics and manufacturing was ignored by too many people in the West yet I gave Andy Grove's 2010 Bloomberg article to many people. The Hyperscalers design chips and even develop EDA software. The history of tech does have lessons for today and so thanks again for this.
Great write-up. I've always wondered how such a dominant company came about. In the end it was a lot of disappointment, great foresight and an investor base who was willing to take a calculated risk.
Very nice work. Morris Chang has overcome failures to turn TSMC into a crown jewel is the net outcome.
Viewing why they struggled so much then while enjoying the fruits of their hardwork is very difficult. It’s like asking a Mango tree why the fruits are offered in summer alone.
Life must’ve taught Morris patience, while he persevered to give us masterpiece
Been a fan of your work for almost two years now, I think. This article was an absolute treat. Thank you! I’ll be sharing this at work. Reminded me of all the semiconductor and fab courses I took in University.
Probably already known but I believe that Emergent Ventures recently awarded a grant to translate Chang’s autobiography.
Incredible service to have translated and summarized this. Thank you.
Really interesting story for thinking about the origins of technical excellence. Even the fact that more recent advances seem uninteresting to him says a lot about how these improvements (at least in certain areas) may actually be very procedural and matter of fact as long as you set up the generating process in the right way.
Thanks again. Great read for heading into the new year.
Amazing story. This man fails at least three times and bounces back. A lesson for young entrepreneurs
Did they cover the near-death of TSMC after Morris retired the first time, they screwed up the gate material transition badly, and he booted the CEO and took over again to fix it?
Roughly 2006 IIRC. It was entirely as bad as Intel's 10nm but Morris reacted swiftly and correcly, while Intel dragged it out. It would be really important for him to cover that era. As well as the gap where he originally parachuted into TSMC and replaced its original CEO.
So good!
I've wanted to read Morris Chang's bio for a long time, and was disappointed to find it was chinese only.
I also enjoyed this conversation (https://youtu.be/wEh3ZgbvBrE?si=alv1Y0xU-sc74uNA) between Morris Chang and former Stanford President John Hennessey. Jensen Huang introduces Chang!
Another interesting tidbit - in an interview Chang was asked why the foundries in Japan didn't take off, and he said it was because they hadn't developed the fabless industry enough to help foundries get to the scale need. I think about that quite a bit seeing how India, America are pushing for onshore semiconductor manufacturing.
In the mid to late 1970s, there was a lot of talk about fabrication as a service. DARPA was pushing its VLSI initiative, and there were a number of novel processor designs being explored. I had some friends, for example, at Symbolics which was manufacturing a LISP machine, and they were designing their own specialty chips. They arranged, probably with some DARPA influence, to get their chips fabricated. It was generally assumed that this was the future of chip manufacturing. There would be design companies and fabrication companies.
As it turned out, this model didn't really catch on. Motorola, Intel and the other big chip companies wanted to control their own fabrication. The US companies were big enough to do this. There were and still are fabricators in the US and elsewhere who will make chips to order, but relying on them meant obeying their design rules and constraints. For example, Bunnie Huang noted that the power regulation tends to be strictly black box. Designers are given a rectangle and power specifications. That's it. Any designer who wanted to explore the power envelope, perhaps to compete with Apple, would be out of luck.
P.S. Phillips is a surprisingly innovative company. They don't get a lot of press, but they did some of the first hook-up-your-TV video games and the first video disk player. Neither product really took off, and they never really dominated any particular market. Despite this, they're still around and still doing innovative stuff. As amazing as the mechanics of EUV chip design may be, the really amazing thing is that a modern corporation took up the challenge, spent the money for years and got the process working. That's almost counter to everything modern capitalism stands for. Is there a blow by blow, perhaps written in Dutch?
Amazing, thanks for doing this. The paragraph about Ford coming within $1 of hiring Chang is crazy. I'm reading Fluke by Brian Klaas about the under-appreciated role of chance and this paragraph could have been a great example. I wonder how Chang would have applied himself to the US auto industry...
Thank you for a very helpful summary. I'm reading The Nvidia Way book (Taiwanese author) that is in English. I joined TI in 1984 and so I remember people like J Fred Bucy. I also recall the success of the Houston DSP design team and the culture of the organisation (I was doing LAN chip design when Wally Rhines ran the micros business). The calibre of people there was exceptional and training better than anywhere. Here in the UK my former colleague Warren East has just been given a knighthood - he was CEO of Arm and RollsRoyce after 12 years at TI. Arm's fabless model led to numerous others that have been acquired. Sir Peter Bonfield is also here and is still on the TSMC Board. The industry has always been global and so I'm interested in the work in Japan and South Korea to match Taiwan. The importance of electronics and manufacturing was ignored by too many people in the West yet I gave Andy Grove's 2010 Bloomberg article to many people. The Hyperscalers design chips and even develop EDA software. The history of tech does have lessons for today and so thanks again for this.
Fascinating !
Really enjoyed this post Brian. 😁
Great write-up. I've always wondered how such a dominant company came about. In the end it was a lot of disappointment, great foresight and an investor base who was willing to take a calculated risk.
Loved it! Thank you very much!
Sometimes business stagger to sucess....or really survive to sell another day...although TSMC had huge growth.
Very nice work. Morris Chang has overcome failures to turn TSMC into a crown jewel is the net outcome.
Viewing why they struggled so much then while enjoying the fruits of their hardwork is very difficult. It’s like asking a Mango tree why the fruits are offered in summer alone.
Life must’ve taught Morris patience, while he persevered to give us masterpiece
Been a fan of your work for almost two years now, I think. This article was an absolute treat. Thank you! I’ll be sharing this at work. Reminded me of all the semiconductor and fab courses I took in University.
Very interesting! Two questions:
1) Do you have any idea what the yield for TSMC is today. And how good is that compared to Samsung and Intel?
2) How many fabs does TSMC currently have?