KEYNOTE: AI Needs RF: Power, Edge Intelligence, and the Role of RF Hardware

As artificial intelligence moves toward the edge, its physical realization is increasingly constrained by power, latency, and system efficiency. While local processing reduces reliance on centralized compute, the high energy consumption of modern AI hardware and limited battery scaling fundamentally reshape system-level trade-offs. This keynote argues that, rather than reducing the importance of communication, edge intelligence re-emphasizes the need for fast, efficient, and adaptable RF links between edge devices and datacenters. It highlights how emerging RF front-end requirements place new demands on selectivity, integration, and reconfigurability, and how advances in acoustic wave technologies — spanning mmWave operation, mode engineering, and reconfigurable architectures — are transforming RF components into flexible building blocks for distributed AI systems.