Making data centers flexible so they can serve the grid rather than stress it out

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.volts.wtf/subscribeTypically, AI data centers are large, inflexible loads that the grid has to build around, which is one reason utilities take so long to connect them. Emerald AI has designed a “digital brain” that can ramp down, move, or delay computing jobs in a data center on demand, making a data center a flexible asset to the grid. I talk with CEO Varun Sivaram about how the software works, what flexibility costs the compute, why it beats just installing batteries, whether utilities can enforce it, and what it would mean for clean energy.Chapters:00:00 – Introduction03:05 – Temporal, spatial, and resource flexibility06:26 – What Emerald Conductor touches on site08:35 – Who decides which workloads can flex10:55 – Who is liable when a job slows down13:45 – Who actually signs the contract14:49 – Larger and faster grid connections18:02 – Enforcing the flexibility promise19:05 – What broke in the demos, from bad nodes to slow telemetry26:41 – What flexing costs the compute jobs29:21 – Why not just use batteries, and the demand merit order35:44 – Training, inference, and substation-scale data centers40:23 – PJM, ERCOT, and legal enforceability44:50 – Renewables, gas, and consumer bills