The mechanics of data center flexibility

In this episode of Catalyst, Shayle Kann interviews Varun Sivaram, CEO of Emerald AI, about data center flexibility and its implications for the electricity grid. They discuss the current state of AI data center electricity consumption, highlighting its unpredictable and energy-dense nature, and how grid operators plan for worst-case scenarios. Varun explains how Emerald AI's software platform aims to make data centers flexible by orchestrating computational workloads, taking advantage of inherent flexibility in AI use cases. They explore temporal and spatial flexibility, the importance of SLAs, and the potential for data centers to become a key asset in enabling a cleaner, more intermittent grid. The conversation also covers the need for demonstrations to build trust with grid operators and the development of digital twins to simulate orchestration operations.