Abstract

The rapid growth of generative AI has driven AI chips to reach unprecedented power and heat densities, pushing traditional air-cooling toward its physical limits. This talk introduces the major cooling approaches—including advanced air cooling, cold-plate liquid cooling, and immersion cooling—and explains their impact on data-center energy efficiency and PUE requirements. The presentation will also share ITRI’s recent experimental results on water-immersion cooling, highlighting future trends in AI thermal management and data-center infrastructure.

Bio.

Tsai-Liang (Terry) Wu is an engineer at the Industrial Technology Research Institute (ITRI), focusing on thermal management and liquid-cooling technologies for next-generation AI chips and data centers. Holding a Master’s degree in Engineering Science from National Cheng Kung University. Work spans high-power chip cooling, nanofluids, advanced thermal materials, and energy-efficient cooling solutions. Experienced in international R&D collaborations and currently engaged in developing and validating water-immersion cooling technologies to support rapidly growing HPC and AI infrastructure.

All faculty and students are welcome to join.