This article examines the evolution of architectural approaches to ensuring energy-efficient computing in edge artificial intelligence systems operating under stringent resource constraints. The aim of the study is to analytically trace the transition from model-centric optimizations to multi-layer energy budget management that encompasses inference metrics, memory organization, data movement schemes, and connectivity with remote infrastructure. The relevance of the work is driven by the rapid growth in the number of edge devices, tightening requirements for latency, privacy, and autonomy, and the emergence of on-device generative models. The novelty of the article lies in a comprehensive reconstruction of the trajectory of architectural development, from compact neural networks, compression, and adaptive execution to hardware–software co-design, hybrid edge–cloud schemes, and the integration of event-driven and neuromorphic principles. It is shown that the key bottleneck is not arithmetic but memory and data transfers, while promising directions include energy-aware compilation and automated design under a given energy budget. The article will be of interest to researchers and engineers involved in the development of edge AI systems and energy-efficient computing architectures.

edge artificial intelligence, energy efficiency, computing architecture, compact neural networks

R. Singh and S. S. Gill, “Edge AI: A survey,” Internet of Things and Cyber-Physical Systems, vol. 3, pp. 71–92, Mar. 2023, doi: https://doi.org/10.1016/j.iotcps.2023.02.004.

W. Raza, A. Osman, F. Ferrini, and F. D. Natale, “Energy-Efficient Inference on the Edge Exploiting TinyML Capabilities for UAVs,” Drones, vol. 5, no. 4, p. 127, Oct. 2021, doi: https://doi.org/10.3390/drones5040127.

T. Wang, J. Guo, B. Zhang, G. Yang, and D. Li, “Deploying AI on Edge: Advancement and Challenges in Edge Intelligence,” Mathematics, vol. 13, no. 11, p. 1878, Jun. 2025, doi: https://doi.org/10.3390/math13111878.

D. Ngo, H.-C. Park, and B. Kang, “Edge Intelligence: A Review of Deep Neural Network Inference in Resource-Limited Environments,” Electronics, vol. 14, no. 12, p. 2495, Jun. 2025, doi: https://doi.org/10.3390/electronics14122495.

Raha, D. A. Mathaikutty, S. Kundu, and S. K. Ghosh, “FlexNPU: a dataflow-aware flexible deep learning accelerator for energy-efficient edge devices,” Frontiers in High Performance Computing, vol. 3, Jun. 2025, doi: https://doi.org/10.3389/fhpcp.2025.1570210.

S. Heydari and Q. H. Mahmoud, “Tiny Machine Learning and On-Device Inference: A Survey of Applications, Challenges, and Future Directions,” Sensors, vol. 25, no. 10, p. 3191, May 2025, doi: https://doi.org/10.3390/s25103191.