Networking-Aware Energy Efficiency in Agentic AI Inference: A Survey

TL;DR

This survey reviews energy efficiency challenges in agentic AI, emphasizing computational and communication costs, and proposes a framework and taxonomy for optimizing energy use in autonomous systems.

Contribution

It introduces an energy accounting framework and a unified taxonomy for energy-efficient agentic AI, highlighting cross-layer co-design strategies and future research challenges.

Findings

Identifies computational and communication energy bottlenecks in agentic AI.

Proposes a unified taxonomy for energy optimization strategies.

Outlines open challenges and a roadmap for scalable autonomous intelligence.

Abstract

The rapid emergence of Large Language Models (LLMs) has catalyzed Agentic artificial intelligence (AI), autonomous systems integrating perception, reasoning, and action into closed-loop pipelines for continuous adaptation. While unlocking transformative applications in mobile edge computing, autonomous systems, and next-generation wireless networks, this paradigm creates fundamental energy challenges through iterative inference and persistent data exchange. Unlike traditional AI where bottlenecks are computational Floating Point Operations (FLOPs), Agentic AI faces compounding computational and communication energy costs. In this survey, we propose an energy accounting framework identifying computational and communication costs across the Perception-Reasoning-Action cycle. We establish a unified taxonomy spanning model simplification, computation control, input and attention optimization, and hardware-aware inference. We explore cross-layer co-design strategies jointly optimizing model parameters, wireless transmissions, and edge resources. Finally, we identify open challenges of federated green learning, carbon-aware agency, 6th generation mobile communication (6G)-native Agentic AI, and self-sustaining systems, providing a roadmap for scalable autonomous intelligence.