TSMC Chairman C.C. Wei (魏哲家) received an honorary doctorate from Asia University on its 25th anniversary, using the occasion to deliver a wide-ranging analysis of how AI is moving beyond data centers into eldercare settings, as well as TSMC's role in the next wave of AI applications.
Why a Book About Corporate Failure Shaped a CEO's Leadership Style
Wei opened with characteristic self-deprecation, telling the audience: "I actually have some regrets." He then quipped that executives who enjoy giving speeches everywhere, or who accept honorary degrees, are almost certainly arrogant — drawing immediate laughter.
He traced the remark to a gift he received in 2018 when he assumed the CEO role: Jim Collins' management classic How the Mighty Fall. The book warns that arrogance is typically the first stage of institutional decline, Wei noted. He credited TSMC's success primarily to founder Morris Chang (張忠謀) and the company's workforce, saying his own contribution "has not been great."
（Related：
Not just GPUs: LinkCom eyes H2 silicon photonics, LEO satellite wins amid AI buildout
｜
Latest
）
Wei said two reasons ultimately changed his mind about accepting the degree. First, he did not want to sever ties with the university's affiliated physicians — joking that he could not afford to "burn that bridge. Second — and more candidly — he noted that opportunities to earn a doctorate "without coursework or examinations" are rare in any lifetime.
How Semiconductor Scaling Made AI Commercially Viable
Wei argued that AI's commercial emergence is inseparable from decades of semiconductor process advancement. When he joined TSMC in 1998, the leading process node was 0.25 microns. Production is now moving toward 2-nanometer chips — a roughly 100-fold improvement in roughly 25 years.
To illustrate the scale of that progress, Wei offered two analogies. If running 100 meters took 10 seconds 20 years ago, an equivalent rate of improvement would reduce that to 0.1 seconds today. If a car cost approximately USD 33,000 two decades ago, proportional progress would bring the price down to roughly USD 330.
Eldercare Robotics: The Next AI Frontier, and Why Semiconductors Are Central
Wei first used AI-assisted medical diagnosis as an entry point. He argued that once AI accumulates sufficient pathological data, regional epidemiological patterns, and clinical diagnostic experience, its accuracy in diagnosing conditions such as influenza could reach approximately 95 percent. Crucially, however, Wei framed this not as the destination but as a starting point: AI assistance in clinical settings merely scratches the surface. The far more urgent and expansive opportunity, he argued, lies in eldercare robotics — a market driven by the structural pressures of an aging society.
The section of Wei's remarks that drew the most external attention concerned what he described as the structural case for eldercare robotics. He was direct: the robots required for aging-society care are not demonstration machines that "do flips and dance." They are complex sensing systems with demanding technical requirements.
Wei outlined three core functional requirements for such robots. The first is vision and tactile sensing — robots must identify spatial positions using optical sensors and calibrate physical contact through pressure sensors to ensure safe interaction. The second is precise temperature sensing and control, enabling the robot to distinguish between, for example, 20 degrees Celsius and 100 degrees Celsius. The third is high-speed connectivity, analogous to neural transmission, through which all sensory data must be relayed via 5G or 6G networks to a central processing unit before being converted into physical action.
（Related：
Not just GPUs: LinkCom eyes H2 silicon photonics, LEO satellite wins amid AI buildout
｜
Latest
）
Wei stated that while the software "brain" driving these systems is largely being developed by major American technology companies, more than 95 percent of the manufacturing originates from TSMC. The claim positions TSMC not merely as a supplier to AI compute platforms such as those produced by Nvidia and AMD, but as an indispensable foundation for the broader sensor and communications components that robotics requires.
The Commercial Threshold: Cost and Reliability
For eldercare robots to enter households at scale, Wei identified two decisive variables: cost control and reliability. He said he personally hopes that any robot he uses in the future will contain transistors manufactured by TSMC — adding that if they came from a competitor, "I would have to think twice."
On AI's role in healthcare more broadly, Wei argued that the technology will not replace physicians but will reduce the burden of repetitive diagnostic tasks. He noted that with sufficient accumulation of pathological data, regional epidemiological patterns, and clinical diagnostic experience, AI-assisted influenza diagnosis could reach an accuracy rate of approximately 95 percent.
Wei also addressed AI in healthcare, saying AI would not replace physicians but would free them from repetitive tasks. Drawing laughter from the audience, he noted that TSMC is often criticized for overworking its employees — yet doctors, exempt from Taiwan's Labor Standards Act, work under even more grueling conditions, a situation he drily described as making them the "lucky" ones. He added that TSMC's ranking as seventh in the country for labor inspection fines was "a bit embarrassing.
Closing the speech, Wei directly addressed students in the audience, encouraging them to join TSMC after graduation. He then offered thanks to his family, colleagues, and founder Morris Chang, saying the support he had received along the way was "absolutely not to be forgotten."
You've read it. Now let's talk. Follow us on X. Editor: Yuping Chang
