TSMC COO Y.J. Mii stated that while artificial intelligence enhances chip design and software efficiency, it cannot replace human engineers in developing next-generation manufacturing processes. Mii compared current AI to a three-year-old superman, noting its power but lack of mature judgment for critical decision-making.
TSMC is currently developing A14 (1.4nm) and A12 process nodes. While AI tools from companies like Synopsys and Cadence assist in optimizing complex circuits with billions of transistors, Mii emphasized that these tasks rely on existing rules and historical data. Creating new nodes involves unknown variables and a lack of foundational experimental data.
Mii argued that AI cannot overcome physical limitations or discover new scientific laws without sufficient data. Innovation in advanced manufacturing requires human intuition and years of engineering practice. AI serves as a tool for efficiency rather than a substitute for core research and development.
Beyond technical challenges, TSMC faces difficulties in international expansion, specifically regarding the recruitment and training of senior management capable of overseeing advanced manufacturing systems. The company currently uses AI to monitor factory construction progress and improve safety management.
