[Presentations (Moderator: Prof. Yoon-Jong Lee, KAIST)] The Role of Semiconductor Packaging in the AI Era: Innovations in Advanced Packaging Technologies Accelerating the AI Era
Generative AI must process and infer from massive volumes of data in real time. Even with highly capable GPUs, performance bottlenecks arise if the data pathway (bandwidth) is constrained. Therefore, scaling AI models and computation requires breakthroughs in data throughput. AI accelerators enabled by 2.5D SiP (System-in-Package) and HBM (High Bandwidth Memory) have been key to this shift. With advantages in power efficiency and thermal management, AI accelerators are now regarded as essential semiconductors for AI data centers.
As core building blocks for high-performance AI systems, 2.5D SiP and HBM—critical for large-scale data computation—are the result of integrated innovations in packaging architecture, processes, and materials. Going forward, continued advances in advanced semiconductor packaging are expected to accelerate the AI era and drive growth across the AI industry.
This talk will review the development of advanced packaging technologies such as HBM and 2.5D SiP that have enabled the AI era, and discuss future directions needed to sustain and lead AI industry growth.