Volume 3 Issue 2, February 2026

Data centres powering artificial intelligence now consume vast amounts of electricity, which raises new sustainability concerns. Underwater data centres powered by offshore wind, solar and wave energy, and cooled by seawater systems, offer a route toward zero-carbon artificial intelligence.

There are challenges in making large-scale and full utilization of renewable energy in power systems. We consider the need to integrate prediction, analysis, dispatch and control — and call for a systems engineering approach to power system operation that can enhance the utilization of renewable energy.

Sang Il Seok talks to Nature Reviews Electrical Engineering about research and education at the School of Energy and Chemical Engineering at Ulsan National Institute of Science and Technology (UNIST), including work to advance perovskite solar cells by pioneering inorganic–organic hybrid heterojunction solar cells that contain perovskite compound and polymeric hole conductors.

A study in Nature Communications introduces a personalized federated learning framework that enables privacy-preserving, high-accuracy detection of battery faults across heterogeneous electric-vehicle charging data.

Researchers introduce sodium sulfamate as a bromine scavenger for zinc/bromine flow batteries, reducing levels of corrosive free bromine. This innovation boosts energy density, cycle life, and corrosion resistance, offering a promising route to stable, cost-effective large-scale energy storage.

This Review provides a holistic view of sustainable gas, liquid and solid insulating materials, focusing on substitution, benign degradation, recycling, reuse and resource conversion to support net zero power systems.

The transition to low-carbon energy resources necessitates resilient power grids capable of handling low-inertia system dynamics. This Perspective proposes an Internet-inspired power system set-up composed of independent, asynchronous compartments able to balance energy across the entire grid.

Decommissioned power semiconductor devices often still have useful life, creating opportunities to reuse rather than discard them. This Perspective outlines how careful screening and creative upcycling, including artificial intelligence-assisted assessment and circuit reconstruction, can extend device lifetimes while reducing waste and supporting a circular and lower carbon power electronics industry.

This Perspective examines the potential of large language models (LLMs) in sixth-generation radiocommunications (6G) networks, highlighting cloud LLM fundamentals, telecom-specific adaptations, edge-device implementation techniques and multi-agent deployments.