AI Infrastructure: From Digital Tokens to Physical Embodied Intelligence

Join us for our upcoming Future Computing Seminar Series

Speaker: Prof. Yu Wang, Tsinghua University

Date: Friday, Sept 11, 10:30 Zurich time (CEST)

Where: external page Livestream on YouTube & Room: HG E 1.1

Abstract:

The rapid growth of foundation models and AI agents is turning AI infrastructure into a “token factory”. Our key objective is to deliver intelligence at the lowest cost, energy, and latency per token. This talk will discuss how end-to-end hardware–software co-design can improve the efficiency, scalability, and reliability of AI systems. I will first present our work on heterogeneous large-scale training and inference, including heterogeneous computing resource management, disaggregated prefill-decode serving, and automated kernel optimization for diverse hardware platforms. As AI moves beyond generating and reasoning over digital tokens toward perceiving, deciding, and acting in the physical world. I will then discuss the emerging infrastructure requirements of embodied intelligence. I will introduce our proposed embodied AI infrastructure that spans simulation, real-world data collection, reinforcement-learning training, cloud-edge execution, and robot-side inference. This talk will conclude with our perspective on AI infrastructure that can continuously evolve with models, agents, hardware, and real-world applications.

Speaker Bio:

Yu Wang is a Professor in the Department of Electronic Engineering at Tsinghua University and an IEEE Fellow. He is a recipient of the National Science Fund for Distinguished Young Scholars (NSFC) and serves as Director of the Beijing Key Laboratory of Heterogeneous Interconnected Sustainable Intelligent Computing Power Chips and Systems. His research interests include application-specific heterogeneous computing, processing-in-memory, intelligent multi-agent systems, and power- and reliability-aware system design. He has published more than 90 journal papers, including 64 in IEEE/ACM journals, and 270 conference papers in EDA, FPGA, VLSI design, and embedded systems. His work has received more than 34,000 Google Scholar citations, as well as five best-paper awards and thirteen best-paper nominations. He co-founded DeePhi Tech, a deep-learning solutions company acquired by Xilinx in 2018, and has also helped drive Infinigence AI Tech, an AI-infrastructure company delivering high-performance large-language-model inference across more than ten chip platforms.