令和7年度 第2回CMDS先端AIセミナー
- タイトル
- Power Play: Harnessing Data Centers Spatial Workload Flexibility
- 講演者
- Prof. Yihsu Chen
Department of Electrical and Computer Engineering
Jack Baskin School of Engineering
University of California, Santa Cruz
- 日程
- 3/21 15:00ー16:00
- 形式
- ハイブリッド開催(瀧川記念学術交流会館大会議室、zoom)
https://www.kobe-u.ac.jp/ja/about/outline/facilities/takigawa/
- 概要
- The rapid adoption of artificial intelligence (AI) has ushered in an era of computational demand, with large language models (LLMs) at the forefront since 2022. This unprecedented growth in loads, induced by the computational demand of data centers, poses daunting challenges to the electric power system. One common belief is that the flexibility provided by data center operations might attenuate these impacts by spatially migrating workloads to modular data centers (MDCs), which are typically located at the network edge. We apply a power market model that considers the problems faced by a hyperscaler, MDCs, and the grid operator where the hyperscaler enters a capacity leasing agreement with MDCs in order to outsource its workload. At the same time, the hyperscaler minimizes operating costs while accounting for associated emissions. We demonstrate that direct load reduction via workload migration at the hyperscaler's host bus can effectively lower local power prices. However, this may inadvertently inflate prices at other locations, ultimately increasing total consumer load payments across the network. The increase in average load payments is positively correlated with the total grid load. Furthermore, it can also exacerbate price volatility, necessitating additional ancillary service procurement to ensure stable grid operations.
- 申込
- https://forms.gle/785k6cWHJKGGCZDi8
(懇親会参加の方は3/14までにご登録ください。講演参加のみの方は当日も受け付けております。)
