Computing at the Edge Enabled by Indoor Photovoltaics and Ferroelectrics

Abstract

The exponential, pervasive rise of artificial intelligence (AI), hosted in large data centres, will test the limits of the underpinning infrastructure, including electricity and water supplies, which are becoming increasingly precious resources. Rather than performing all AI computations in centralized servers, for many functionalities, these can be performed locally across a network of billions of small, autonomous nodes, saving substantial energy costs associated with communication. This alternative paradigm is known as edge computing, or distributed intelligence (DI), but has been held back by the lack of availability of reliable local energy supplies matching the energy requirements of the computational infrastructure. Opportunities to address this challenge are emerging with rapid advances in high-performance indoor photovoltaics (IPVs) for local energy harvesting, as well as reductions in computational cost through neuromorphic or in-memory computing devices. In this perspective, we discuss the requirements of IPVs for DI, the extent to which emerging materials fulfil these requirements, and how current gaps could be addressed. We make the case that (anti)ferroelectric materials can surpass bottlenecks for both electrostatic energy storage and low-power in-memory neuromorphic computing. The core theme of this perspective is that co-creation between energy harvesting, storage and computing is essential for making DI a reliable and more sustainable alternative to centralized AI.