3Dx3D System Scaling for Ubiquitous and Zetta-scale Compute/AI
The rapid development of artificial intelligence, including generative models such as ChatGPT, places significant demands on computational performance and energy efficiency. Today’s models operate with trillions of parameters, and their computational requirements are growing at a pace that exceeds the capabilities of conventional semiconductor scaling.
The lecture will introduce the concept of 3Dx3D system scaling, combining 3D CMOS and 3D heterogeneous integration, which underpins current advances in AI, and will open a discussion on the need for a fundamental shift towards new computing architectures.
The talk will address the need for dramatically improved energy efficiency, new approaches to system, algorithm, and materials design, and the challenges associated with further scaling of AI. It will also offer a broader perspective on the evolution of semiconductor technologies and their role in building a sustainable future.
About the Speaker
Jack Y.-C. Sun is a renowned expert with more than 40 years of experience in semiconductor research and development. He previously served as CTO at TSMC and held leading research positions at IBM. He has made significant contributions to the development of CMOS technologies, 3D transistor structures, and the concept of 3Dx3D system scaling.
He currently serves as Dean of the Industry-Academia Innovation School and Senior Vice President at National Yang Ming Chiao Tung University.
