Breakfast Bytes Blogs

A couple of weeks ago, Morris Chang, the founder of TSMC and, for a long time (twice), its CEO, was interviewed at the Brookings Institution and the Center for Strategic and International Studies (CSIS), two think tanks in Washington DC. The topic was Can Semiconductor Manufacturing Return to the US?

The interview starts with some of Morris's background. He was born in pre-communist China, but when turmoil increased, his family fled to Hong Kong. He had just finished high school and had started at a university in Shanghai. Since he didn't want to attend a university in Hong Kong, and he didn't want to go back to China, he came to the US. For much of his career, he worked at Texas Instruments and rose to manage the whole TI semiconductor business. He thought that TI should never have gotten into the consumer products business (calculators, speak'n'spell, etc.), and so when he was put in charge of it, he considered it a downgrade and left the company. After a couple of other positions, Taiwan called. They wanted him to run ITRI, the very generically named Industrial Technology Research Institute.

Morris realized that ITRI did good research, but without the pressures of a running business, they were largely spinning their wheels. Instead, he took the team of researchers and created a company to house them. That company was TSMC. By the way, he was not the first CEO, just as Elon Musk was not Tesla's first CEO. See my post TSMC 30 Years Ago Today. To some extent, the reason that TSMC is a pure-play foundry, he explained, is that Taiwan had no design expertise, They had manufacturing but were several process generations behind the leaders, although the yield was very good. So doing manufacturing but not design was a perfect fit for Taiwan.

By being a foundry, we skipped over our weaknesses and used our strengths. A foundry did not need design strengths. While the lack of advanced nodes is a weakness for a foundry, there is a lot of foundry business on the mature nodes. Everyone still wanted the old products, but TI and Intel and Motorola were so busy trying to advance their nodes that they really didn't care about the mature nodes, but they still sold a lot of mature node products. If a foundry was willing to do the mature nodes for them, and had advantageous cost, then they would be very happy to let the foundry do them. And TSMC fulfilled that role.

A theme of the interview is that for the US, it is the other way around, to do design but not manufacturing.

The US has a ready supply of design talent. It's the best in the world. Taiwan has very little design talent, and TSMC has absolutely none.

Back in the fifties, sixties, and the early part of the seventies, the US had these manufacturing talents. But then, starting in the seventies, the manufacturing talent migrated to higher paying professions. To design, if they stayed in the semiconductor industry, or to the internet industry, or finance.

There's a lack of manufacturing talent. I don't think that's a bad thing for the United States but it is a bad thing for trying to bring up semiconductor manufacturing in the US.

I don't think I actually knew that TSMC had a plant in Oregon, but they have had one for 25 years. That plant has given the company knowledge of costs in the US. Morris said:

Costs for manufacturing in the US are simply prohibitive, and TSMC has the data to prove it, thanks to 25 years of manufacturing at its plant in Oregon.
...
We were extremely naive in expecting comparable costs, but manufacturing chips in Oregon, the same product, is about 50 percent more expensive than in Taiwan. It is still profitable, but not nearly as profitable as the Taiwan product, so we have maintained it.

Arizona will be much bigger, much bigger scale, much more advanced technology.

His view on the question we started with is summed up in a single sentence:

We think that the recent efforts of the US to increase onshore manufacturing of semiconductors...right now, you are talking about only spending tens of billions of dollars...it's not going to be enough. I think it will be a very expensive exercise in futility.

The Interview

Listen to the whole interview. Note that although this is a YouTube video, there is actually no video beyond the title card. The first time I listened to it as a podcast, if you prefer your interview that way.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.

.