AI performance gains hinge on materials, not chips—Geckos chair

The article presents an interview-based perspective on a strategic inflection point in semiconductor competition. For decades, the industry focused on Moore's Law—the race to smaller process nodes—as the primary driver of chip performance. However, as generative AI workloads push demand for computing power and process nodes approach physical and economic limits (2nm and below), materials science is emerging as the next performance frontier. Geckos chairman Shen Tsung-huan positions materials innovation as the decisive factor for the next generation of AI chips. This view reflects a broader industry recognition that incremental process improvements yield diminishing returns, while breakthrough materials—whether in interconnects, dielectrics, or other components—could unlock substantial performance and efficiency gains. For chipmakers, chip designers, and materials suppliers, this signals where research investment and competitive differentiation will concentrate in the coming years.