Our country's semiconductor industry is facing a significant challenge. In the rapidly evolving AI supercomputing market led by NVIDIA and TSMC, companies like Samsung Electronics and SK Hynix are primarily relegated to the role of simple suppliers of memory chips or HBM (high-bandwidth memory) packages. Compared to their prominent position in the CPU era, their influence has notably diminished, and no clear strategic vision has emerged to reclaim leadership in the industry.

In contrast, Google has leveraged its groundbreaking experiences with AlphaGo and AlphaFold 2 (the latter contributing to a Nobel Prize in Chemistry this year) to automate computer chip design through the AlphaChip project, revealed in a 2020 Nature paper.

This week’s Data Science seminar aims to explore new ideas to address the diminished status of Korea’s semiconductor industry. We have invited Dr. Igor Markov, a Distinguished Architect from Synopsys (the leading semiconductor CAD tools company), who has engaged in detailed debates with Google’s AlphaChip team. Dr. Markov, who previously worked at Google, has raised concerns about the academic integrity of Google AlphaChip’s Nature paper through the Communications of the ACM journal.

Key Details:

• Title: Challenges and Opportunities for AI in Commercial Chip Implementation Tools
• Speaker: Dr. Igor Markov
• Date and Time: December 4, 2024 (Wednesday), 09:00–10:30
• Platform: Zoom (https://snu-ac-kr.zoom.us/j/2507466555)

Relevant Resources:

1. Dr. Markov’s YouTube video and CACM article:
   • YouTube
   • CACM Article
2. Google AlphaChip’s Nature paper and related discussions:
   • Google Blog Post
   • PubPeer Critique
3. Google’s rebuttal (co-authored by Jeff Dean):
   • arXiv

Through Dr. Markov’s lecture, we aim to gain insights into how combining AI and semiconductor design domain expertise can pave the way for the future development of the semiconductor industry.