TSMC’s $100bn pledge to Donald Trump will not revive US chipmaking, says ex-Intel chief

[hist78]

"Taiwan Semiconductor Manufacturing Company’s pledge to spend an extra $100bn on advanced manufacturing plants in the US will do little to help the country restore its global lead in chipmaking, according to Pat Gelsinger, who was forced out as chief executive of Intel late last year.

His comments come less than a month after the White House hailed the investment from TSMC, the world’s largest chip manufacturer, as an important milestone in efforts to bring production of the most advanced semiconductors back on to US soil.

“If you don’t have R&D in the US, you will not have semiconductor leadership in the US,” Gelsinger said. “All of the R&D work of TSMC is in Taiwan, and they haven’t made any announcements to move that.”

The former Intel chief added, however, that President Donald Trump’s tariff threats had at least been “incrementally beneficial” for the US by giving chip manufacturers such as TSMC more incentive to locate their facilities in the country.

The Trump administration has leaned heavily on TSMC amid doubts about Intel’s ability to regain the global manufacturing lead it lost to the Taiwanese company a decade ago."

TSMC’s $100bn pledge to Donald Trump will not revive US chipmaking, says ex-Intel chief

Pat Gelsinger, who has taken on role at venture capital firm, says US must boost R&D to gain ‘semiconductor leadership’

www.ft.com

 

[hist78]

"Taiwan Semiconductor Manufacturing Company’s pledge to spend an extra $100bn on advanced manufacturing plants in the US will do little to help the country restore its global lead in chipmaking, according to Pat Gelsinger, who was forced out as chief executive of Intel late last year.

His comments come less than a month after the White House hailed the investment from TSMC, the world’s largest chip manufacturer, as an important milestone in efforts to bring production of the most advanced semiconductors back on to US soil.

“If you don’t have R&D in the US, you will not have semiconductor leadership in the US,” Gelsinger said. “All of the R&D work of TSMC is in Taiwan, and they haven’t made any announcements to move that.”

The former Intel chief added, however, that President Donald Trump’s tariff threats had at least been “incrementally beneficial” for the US by giving chip manufacturers such as TSMC more incentive to locate their facilities in the country.

The Trump administration has leaned heavily on TSMC amid doubts about Intel’s ability to regain the global manufacturing lead it lost to the Taiwanese company a decade ago."

Try open the following link in your browser's private window/tab.

https://www.google.com/url?sa=t&source=web&rct=j&opi=89978449&url=https://www.ft.com/content/68a0da9b-ffc7-4c67-8249-d7c452efa864&ved=2ahUKEwix5Y62sKyMAxVakIkEHRAxEJUQxfQBKAB6BAgGEAE&usg=AOvVaw3Qe6_dUaf9O9ItyE25h9n1

What a party pooper

 

[blueone]

What a party pooper

Gelsinger is correct.

 

[BruceA]

Fab21 P1 is what 20K coming online this year compared to 160K in Tainan.

Fab21 P2 is what maybe another 20K of N3 coming online in 2027 more likely 2028 a good three years after 3nm ramp in Tainan of similar capacity as N5.

Most of that Taiwan capacity is going to US companies not sure how 1/6 of capacity coming online three to four years later makes America supply insecure.

At best the Arizona fab will bring trailing edge 15–20% incremental capacity online three to four years later than the lead ramp in Taiwan, hardly call that leading edge nor security of leading edge.

All thsi talk about learning and accelerating is true but if acceleration from the P1 schedule by even 50% still leaves P2 and P3 three years and more behind Taiwan and finding the necessary labor is still a problem they haven’t solved

 

[tooLongInEDA]

"Taiwan Semiconductor Manufacturing Company’s pledge to spend an extra $100bn on advanced manufacturing plants in the US will do little to help the country restore its global lead in chipmaking, according to Pat Gelsinger, who was forced out as chief executive of Intel late last year.

His comments come less than a month after the White House hailed the investment from TSMC, the world’s largest chip manufacturer, as an important milestone in efforts to bring production of the most advanced semiconductors back on to US soil.

“If you don’t have R&D in the US, you will not have semiconductor leadership in the US,” Gelsinger said. “All of the R&D work of TSMC is in Taiwan, and they haven’t made any announcements to move that.”

The former Intel chief added, however, that President Donald Trump’s tariff threats had at least been “incrementally beneficial” for the US by giving chip manufacturers such as TSMC more incentive to locate their facilities in the country.

The Trump administration has leaned heavily on TSMC amid doubts about Intel’s ability to regain the global manufacturing lead it lost to the Taiwanese company a decade ago."

Try open the following link in your browser's private window/tab.

https://www.google.com/url?sa=t&source=web&rct=j&opi=89978449&url=https://www.ft.com/content/68a0da9b-ffc7-4c67-8249-d7c452efa864&ved=2ahUKEwix5Y62sKyMAxVakIkEHRAxEJUQxfQBKAB6BAgGEAE&usg=AOvVaw3Qe6_dUaf9O9ItyE25h9n1

Hold on - isn't this the same Pat Gelsinger who was telling us how great Intel's technology was for the past few years ? If Intel's already as good as TSMC and on course to repass it (as Pat repeatedly claimed), what does it matter whether TSMC's R&D is in Taiwan or the US ? Pat's whole strategy was based around Intel beating TSMC on process technology leadership. He seems to be complaining that TSMC isn't fixing the problem he claimed he'd already fixed ...

Still, at least in the rather colourless modern era (post Jerry Sanders, TJ Rodgers and co), Pat's a reminder of the semi CEOs who could be relied on for a memorable quote.

 

[fansink]

Hold on - isn't this the same Pat Gelsinger who was telling us how great Intel's technology was for the past few years ? If Intel's already as good as TSMC and on course to repass it (as Pat repeatedly claimed), what does it matter whether TSMC's R&D is in Taiwan or the US ? Pat's whole strategy was based around Intel beating TSMC on process technology leadership. He seems to be complaining that TSMC isn't fixing the problem he claimed he'd already fixed ...

Still, at least in the rather colourless modern era (post Jerry Sanders, TJ Rodgers and co), Pat's a reminder of the semi CEOs who could be relied on for a memorable quote.

Bullseye

I would never label Pat as a blusterer, scapegoater, or charlatan, I’ll leave that to others, but given his claims of being God-fearing, perhaps boastful believer is apropos.

 

[Daniel Nenni]

What a party pooper

Pat G is not wrong about the importance of localized R&D. He is wrong about TSMC however. TSMC already announced establishing an R&D presence in Arizona. Pat must have missed that memo. Replicating R&D in the US, however, should come after the fabs have been built, not before, so TSMC is right on course.

TSMC also collaborates with US Universities and the massive semiconductor ecosystem of partners and customers. I have always said that TSMC's reported R&D budget is many times more than what is reported as a result of these ecosystem R&D collaborations, absolutely.

 

[hist78]

To me, one of the biggest questions is: Where *was* Intel, the self-proclaimed American semiconductor advocate, over the past 10+ years?

Before the upcoming Arizona Fab 52, Intel had not built a high-volume, leading-edge fab on US soil for a decade!

 

[blueone]

Replicating R&D in the US, however, should come after the fabs have been built, not before, so TSMC is right on course.

I'm not following you. Why would anyone "replicate" R&D? That seems illogical to me. Also, why does R&D come after fabs? TSMC has numerous existing fabs. An "R&D presence" could mean almost anything.

 

[Daniel Nenni]

I'm not following you. Why would anyone "replicate" R&D? That seems illogical to me. Also, why does R&D come after fabs? TSMC has numerous existing fabs. An "R&D presence" could mean almost anything.

To support local manufacturing:

HSINCHU, Taiwan, R.O.C. Mar. 4, 2025 – TSMC (TWSE: 2330, NYSE: TSM) today announced its intention to expand its investment in advanced semiconductor manufacturing in the United States by an additional $100 billion. Building on the company’s ongoing $65 billion investment in its advanced semiconductor manufacturing operations in Phoenix, Arizona, TSMC’s total investment in the U.S. is expected to reach US$165 billion. The expansion includes plans for three new fabrication plants, two advanced packaging facilities and a major R&D team center, solidifying this project as the largest single foreign direct investment in U.S. history.

https://pr.tsmc.com/english/news/3210

 

[blueone]

To support local manufacturing:

HSINCHU, Taiwan, R.O.C. Mar. 4, 2025 – TSMC (TWSE: 2330, NYSE: TSM) today announced its intention to expand its investment in advanced semiconductor manufacturing in the United States by an additional $100 billion. Building on the company’s ongoing $65 billion investment in its advanced semiconductor manufacturing operations in Phoenix, Arizona, TSMC’s total investment in the U.S. is expected to reach US$165 billion. The expansion includes plans for three new fabrication plants, two advanced packaging facilities and a major R&D team center, solidifying this project as the largest single foreign direct investment in U.S. history.

https://pr.tsmc.com/english/news/3210

I remember reading that (on this forum), and I still don't understand what it means. Does that mean US fabs and Taiwan fabs will run different processes that need different R&D? Or just another site that is part of the larger TSMC R&D team, just like numerous chip design companies have teams distributed in many countries, with Taiwan directing the work?. I didn't think fab process R&D lent itself to the distributed model like product design does. (And product design has a tenuous relationship with distributed R&D. Distributed chip design and development only really works in my experience when you have partitioned responsibilities, not subservient sub-groups with centralized control.)

 

[fansink]

To support local manufacturing:

HSINCHU, Taiwan, R.O.C. Mar. 4, 2025 – TSMC (TWSE: 2330, NYSE: TSM) today announced its intention to expand its investment in advanced semiconductor manufacturing in the United States by an additional $100 billion. Building on the company’s ongoing $65 billion investment in its advanced semiconductor manufacturing operations in Phoenix, Arizona, TSMC’s total investment in the U.S. is expected to reach US$165 billion. The expansion includes plans for three new fabrication plants, two advanced packaging facilities and a major R&D team center, solidifying this project as the largest single foreign direct investment in U.S. history.

https://pr.tsmc.com/english/news/3210

Further details:

TSMC has 3 primary R&D centers located in Hsinchu, Taiwan (Global R&D Center and Fab 12 R&D component) and Tsukuba, Japan (Japan 3DIC R&D Center), with 1 additional U.S. R&D center planned (location TBD). Additionally, there are 6 satellite R&D sites integrated within production fabs: Fab 14 (Tainan, Taiwan), Fab 15 (Taichung, Taiwan), Fab 18 (Tainan, Taiwan), Fab 20 (Hsinchu, Taiwan), Nanjing Fab (Nanjing, China), including a satellite R&D office at Fab 21 in Arizona.

 

[sonchoy]

To support local manufacturing:

HSINCHU, Taiwan, R.O.C. Mar. 4, 2025 – TSMC (TWSE: 2330, NYSE: TSM) today announced its intention to expand its investment in advanced semiconductor manufacturing in the United States by an additional $100 billion. Building on the company’s ongoing $65 billion investment in its advanced semiconductor manufacturing operations in Phoenix, Arizona, TSMC’s total investment in the U.S. is expected to reach US$165 billion. The expansion includes plans for three new fabrication plants, two advanced packaging facilities and a major R&D team center, solidifying this project as the largest single foreign direct investment in U.S. history.

https://pr.tsmc.com/english/news/3210

Dan, TSMC will NOT be moving leading edge R&D to the US, period.

Here's the interview from Taiwan's representative to the US; see at 3:27; he's crystal clear - "The newest technology will be first made in Taiwan; then eventually will go to other places including the United States."

From C.C. Wei himself -

TSMC CEO stresses most R&D will stay in Taiwan | Taiwan News | Mar. 7, 2025 11:50

CC Wei says TSMC has 10,000 R&D researchers in Taiwan, while US center will have 1,000 | Mar. 7, 2025 11:50

www.taiwannews.com.tw

"When asked whether setting up an R&D center in the US would weaken Taiwan’s R&D competitiveness, Wei said centers in America and Taiwan serve different purposes. He said TSMC’s production lines have a unique characteristic compared to global competitors.

Wei said that after a process technology is developed by its researchers and enters the production stage, the production line team continues to refine and optimize the process.

Wei said the engineers responsible for process improvements are also referred to as R&D personnel, but differ from those who develop new process technologies. For example, Wei said TSMC’s 5nm technology was developed by R&D personnel, while the 4nm process was an optimization by the production line.

Similarly, the 3nm process was developed by R&D personnel, whereas N3P (an enhanced 3nm chip) was refined by the production line. These advancements ensure continuous technological progress, and TSMC refers to this as R&D as well, the CEO said.

Wei added that TSMC’s 1nm chip development is led by its R&D personnel, with around 10,000 researchers in Taiwan. The R&D center in the US will have around 1,000 people, mainly focused on improving production line technologies at its US fabs, said Wei.

The CEO said this allows the US facilities to optimize their processes without needing to seek approval from Taiwan for every decision, which is consistent with TSMC’s standard production line approach."

In short, "Development" stays in Taiwan; "Refine and Optimize" can go elsewhere.

 

[fansink]

Dan, TSMC will NOT be moving leading edge R&D to the US, period.

TSMC already has a "leading edge" (4nm) satellite R&D at Fab 21 in Arizona.

 

[sonchoy]

TSMC already has a "leading edge" (4nm) satellite R&D at Fab 21 in Arizona.

Yeah, but they're not doing "Process Technology Development" in Arizona. That is the critical part. C.C. Wei himself said this and I referred to his statement in my previous post.

Without "Process Technology Development" (Recipe) in the US, what good are the factories (Output)?

 

[fansink]

There are only 3 R&D Centers in the entire world, that's like 1 R&D Center for every 4-5 fabs.

The satellite R&D sites supplement localized process work, while the primary R&D Centers drive broader innovation.

 

[sonchoy]

There are only 3 R&D Centers in the entire world, that's like 1 R&D Center for every 4-5 fabs.

The satellite R&D sites supplement localized process work, while the primary R&D Centers drive broader innovation.

I think you touched upon the core of the issue. If Intel goes under, then there is no more "Primary R&D Center" in the U.S. TSMC will not be moving their "Primary R&D Center" to the U.S. as both the Taiwan Rep and C.C. Wei made clear.

 

[fansink]

Don't worry about TSMC, C.C. Wei is more than capable of determining when and where to build R&D Centers, satellite R&D offices, and fabs.

 

[sonchoy]

Don't worry about TSMC, C.C. Wei is more than capable of determining when and where to build R&D Centers, satellite R&D offices, and fabs.

I am not worried about TSMC. I'm worried about US not having any Process Technology R&D on its soil if Intel goes under. It is a grave risk and a national security issue in my view.

I don't see any reason for C.C. Wei to move his Primary R&D to the U.S. Even if he wanted to, Taiwanese government would likely prevent it as it weakens the Silicon Shield.

It is a bad idea for such a crucial industry for the world to be so dependent on just one company, TSMC.

 

[hskuo]

I'm not following you. Why would anyone "replicate" R&D? That seems illogical to me. Also, why does R&D come after fabs? TSMC has numerous existing fabs. An "R&D presence" could mean almost anything.

There are lots of misunderstanding of the function of tsmc's R&D center in Arizona. Typically, R&D should include "Pathfinding/Research", "New process technology development", "Deployment/Transfer". In Foundry world, sometimes there will be some derived sub-nodes which could be dumb shrink, need some process twists and doing deployment and transfer. It might happen before next main node (N+0.5) or between main node N and N-1 node (N-0.5). tsmc put this part into "Fab R&D" activities which I believe it will be major function for the RD center in Arizona Fab. Do we see tsmc setup this function in China Fab/Japan Fab or the Future Germany Fab? The answer is no now. There is a RD center in Japan for different purpose which I exclude it here.