Exclusive | TSMC’s Chief Scientist H.-S. Philip Wong: Long-Term Containment of China Won’t Work, Taiwan Faces a Talent Crisis
Liang-rong Chen
In May, I met Dr. H.-S. Philip Wong again during a closed-door workshop at Stanford. He is TSMC’s Chief Scientist and a professor of Electrical Engineering at Stanford University. He sat beside me with a backpack and took notes directly on an ultra-thin e-paper tablet with a stylus.
He chuckled as he explained that the device had built-in handwriting recognition, “but my handwriting is too bad for it to recognize, so I just save everything as images.”
This understated and easygoing demeanor was at odds with my first impression of him.
That was in 2019 at SEMICON Taiwan, where Dr. Wong delivered a keynote speech. It was his first public appearance after joining TSMC as Vice President of Corporate Research. I recall it as one of the most brilliant keynotes at the exhibition.
With a commanding voice and fluent English, he confidently addressed industry leaders from various countries and declared, “Moore’s Law is alive and well. We’re on track to reach 0.1 nanometers, the scale of a hydrogen atom.”
Dr. Wong's authoritative address perfectly complemented TSMC’s newly emerged status as a global leader, having just surpassed Intel, earning considerable praise from the industry.
Two years later, Dr. Wong returned to academia as his secondment ended, transitioning to an advisory role at TSMC as Chief Scientist.
Born in Hong Kong, Wong moved to the United States after graduating from the University of Hong Kong, earning his Ph.D. in Electrical Engineering from Lehigh University. Before joining Stanford, he led advanced semiconductor research at IBM.
His personal website states that his research interest is “converting scientific discoveries into practical technologies.” One of his most notable accomplishments was leading students to build the world’s first carbon nanotube computer in 2013.
In our interview on Stanford’s campus, Dr. Wong’s tone was humble, yet he spoke incisively about the challenges in future semiconductor technology, the blind spots in U.S. policies aimed at containing China’s semiconductor industry. He also reflected on his experience building TSMC’s advanced R&D team and participating in the planning of the semiconductor academies.
He emphasized that these were all his personal views and could not represent TSMC. Below is an excerpt from the interview:
Q: Why did you join TSMC in 2018 as Vice President of Corporate Research?
Dr. Wong: I had been teaching for 14 years and felt I was getting further and further away from the technical details. I thought it was time to take a sabbatical—Stanford allows professors to take up to two years off to work in industry—and the moment had come for me to do just that.
At the time, TSMC had just surpassed Intel at the 7nm node, becoming the world’s leading chipmaker. Then Co-CEO Mark Liu（劉德音） invited me to build an entirely new research lab, which I saw as an exciting challenge.
Q: Didn't TSMC already have R&D labs?
Dr. Wong: Of course. But what Mark Liu（劉德音） asked me to focus on was research that looked further ahead. He told me that now that TSMC was a leader, we couldn’t just be a fast follower anymore—waiting to see what others did and then following suit. We had to start exploring technologies not directly linked to the current product roadmap.
This is how I see it: an industry leader must have a forward-looking research team. What they develop may not necessarily lead to mass production, but you need the capability to identify good technologies out there—even if you didn’t create them yourself—and know how to acquire them. Otherwise, you risk missing the boat.
I built a small team drawing talent from universities, other companies, and internally from TSMC, with members based in both Hsinchu and California—about 20 people in San Jose. Close interaction with the external research community was essential. In recent years, we’ve had a solid record of academic publications; previously, TSMC only published work on technologies that had already been commercialized.
Our primary task was to establish a fundamental understanding of materials and semiconductor devices to see if they had any fundamental technical issues.
Then, in the next phase of R&D, the pathfinding team begins integrating various technologies—an area where we truly excel. But before moving into integration, we need to make sure there are no fundamental issues. If basic problems surface during the integration stage, they can quickly become major ones.
Some people believe that spending money on basic research is a luxury and should be left to universities, and we should just read their papers.
But someone needs to be in the game to truly recognize what constitutes valuable research.
For a company like TSMC with hundreds of billions of dollars in revenue, allocating a small portion of the budget as strategic insurance makes sense.
Lithography Is Becoming Less Important
Q: You've said lithography is becoming less important. What do you mean by that?
Dr. Wong: That’s not the mainstream view. I’m not saying it’s unimportant, but it’s becoming less important over the long term, say, 20 years.
Over the past 60 years, lithography has been essential. But today, even the most advanced High-NA EUV tools are too expensive and consume too much power. Our current feature sizes are already tiny. How much smaller do we really need to go? Maybe we just don’t need that kind of extreme resolution anymore.
Instead, we should find ways to reduce the cycle time (time required for each process step).
The overall process has become too time-consuming—fabrication to packaging can now take as long as seven months. (Editor’s note: Five months for leading-edge fabrication, plus two more for CoWoS.) Over the past few decades, we've prioritized resolution gains, which have made process flows increasingly complex. Now with added innovations like backside power delivery and stacked FinFETs, cycle times will only continue to grow.
Q: Is it really possible to reduce cycle time?
Dr. Wong: I think it's just a matter of whether we want to do it or not. As I suggested this to equipment vendors, some said, “We can’t do that; customers will buy fewer machines.”
So, for equipment manufacturers, there’s currently no economic incentive. But I believe that as new technologies like 3D packaging continue to emerge, there will be growing economic incentives.
Long-Term Containment of China Won’t Work
Q: So if the current strategy to contain China’s semiconductor industry is to restrict access to cutting-edge EUV lithography tools, wouldn’t that become ineffective if the future of the technology moves in a different direction?
Dr. Wong: That depends on the time horizon you're considering. Over five years, maybe. Over 20? Probably not.
Competition with China, I believe, will be measured in decades, like Deng Xiaoping’s reforms began in 1978, which was fifty years ago.
Q: So, in your view, are the current export restrictions on China’s semiconductor industry effective?
Dr. Wong: What I’ve seen is that they’ve created a market for the Chinese domestic equipment industry. Even if they managed to develop their own equipment before, they couldn’t sell it because Chinese chipmakers would just buy from American or Japanese companies, whose tools were of higher quality. So why would they choose domestic alternatives? But now, these restrictions have created a market for them.
Q: How do you see the state of semiconductor talent in China? It’s reported that Chinese undergraduates now account for more than half of the global student population in this field, leading some to believe that China could soon develop self-sufficient, homegrown technologies—similar to what we’ve seen with DeepSeek.
Dr. Wong: If you take a long view, 20 years ago, the quality of Chinese research papers was relatively low. But in the past 5 to 10 years, they’ve caught up. In some major conferences, Chinese universities now publish more than any U.S. school.
However, I have observed that Chinese universities are not yet able to initiate new research directions. But once a new research direction is established, they truly do better than anyone else. Just like in manufacturing, they are very good at catching up.
Taiwan Faces a Semiconductor Talent Shortage
Q: What’s your view on Taiwan’s semiconductor talent?
Dr. Wong: While I was working at TSMC, I was often asked to review research grant proposals. One case really shocked me.
During one grant review session, a reviewer asked the applicant, “Has anyone studied this issue before?” The applicant replied, “No, this is a completely new topic.” The reviewer then said, “I can’t fund this project—if no one has ever researched this before, it probably means it’s not a viable research direction.”
That’s the opposite of how things work in the U.S., where original thinking is key to innovation. With that mindset, you’ll never break new ground.
One more concern is that the number of semiconductor PhD students in Taiwan is shrinking, leaving professors with fewer resources to support their research.
Q: In recent years, TSMC has helped establish semiconductor academies at Taiwan’s four leading universities—NTU, NCKU, NTHU, and NCTU—as part of an effort to address the talent shortage. You were involved in those discussions as well, right?
Dr. Wong: The origin of the semiconductor academies is quite interesting. Back when I was at TSMC, I often discussed talent development with Mark Liu（劉德音）. We came up with an idea: to establish an institution dedicated to advanced research—not housed within the company, and not part of a university, but independently operated, much like the Institute for Advanced Study in Princeton. Since Mark Liu（劉德音） frequently met with President Tsai Ing-wen（蔡英文） at the time, the proposal eventually made its way to the Presidential Office.
Q: Another key goal of establishing the semiconductor academies was to bypass Taiwan’s constraints on faculty compensation, so that world-class professors could be recruited. But it seems that goal hasn’t been achieved in recent years?
Dr. Wong: That’s right. I'm not sure of the specific reasons. But I believe that if we want to recruit top talent from abroad, it should be done at a national level, not by individual universities. But now, the semiconductor academies are fragmented across various universities, resulting in no synergy. For example, if someone from Stanford wanted to go to Taiwan, they wouldn't know which semiconductor academy to go to.
Q: So you think these academies shouldn’t be affiliated with any single university?
Dr. Wong: Exactly. Universities should collaborate, but the semiconductor academies should not be affiliated with any single university.
Q: Why did things turn out this way, then?
Dr. Wong: I honestly don’t know.
Read more here:
Can graduates of Taiwan’s semiconductor academies close the talent gap?
Nvidia CEO Calls for Policy Shift on AI Exports
TSMC: ‘Physical AI’ Will Feed Next Wave of Semiconductor Boom
非常有趣！really great work.
Fantastic reporting! Love it!