Nokia Bell Labs Examines AI Energy Efficiency in Networks
Nokia Bell Labs is investigating how to manage the growing energy demands of artificial intelligence (AI) within network technology. Research focuses on smaller models, efficient training, and new hardware architectures.
Nokia Oyj's research arm, Nokia Bell Labs, has initiated a comprehensive examination into the effects of artificial intelligence's (AI) escalating energy consumption on telecommunications networks. The widespread adoption of AI necessitates substantial computing power and energy, raising concerns about its environmental footprint.
The research highlights the critical need for developing more energy-efficient AI models and deployment strategies, particularly within network environments. The objective is to identify solutions that balance the benefits offered by AI against its energy demands. This challenge is compounded by forecasts predicting a doubling of data center electricity consumption by 2030.
Nokia Bell Labs' approach involves several key strategies. These include utilizing smaller, purpose-built AI models, enhancing model training through optimized data usage, and employing event-driven computation. The study also explores brain-inspired architectures like spiking neural networks (SNN) and novel hardware solutions, such as analog chips and photonic circuits, which promise significant reductions in energy use.
The company has published a paper outlining a three-step guide for pursuing more sustainable AI integration in networks. This process begins with assessing the necessity of AI and culminates in the application of optimization techniques and specialized hardware. Through these efforts, Nokia aims to ensure that AI deployment in network infrastructure is conducted sustainably.