Abstract
The expansion of AI and data centers creates interdependent challenges across electricity supply, cooling, controls, and cybersecurity. Addressing these challenges requires coordinated decisions about physical resilience, infrastructure costs, and the evidence needed to support operational and investment commitments.
This lecture presents a product-management framework connecting three questions: how should AI infrastructure withstand disturbances and recover safely; how should capacity, flexibility, and stranded-asset risks be allocated; and what evidence should justify increasing operational autonomy? Drawing on industrial practice, the discussion examines measurable requirements, lifecycle accountability, and the distinction between demonstrated performance and projected benefits.
The lecture concludes with interdisciplinary research opportunities in demand uncertainty, resilience valuation, load-flexibility measurement, and assurance for AI-enabled control. It explores how shared testbeds, reproducible benchmarks, and decision-ready methods can connect academic research with utility, data-center, and industrial operating decisions.
|