This episode argues that physical security needs to move beyond simply adding more powerful GPUs to existing systems. As AI develops from detecting objects and events to actually reasoning, making decisions, and taking action, security systems need a fundamentally different architecture.

Yudhvir Rana of Intel proposes changing the central question from “How much GPU do I need?” to “What decisions do I want the system to make, and where should those decisions happen?