UK scientists to tackle AI's surging energy costs with atom-thin semiconductors

1 min read

A team of UK scientists from Queen Mary University of London, and the universities of Nottingham and Glasgow has received a £6m EPSRC programme grant to develop energy efficient, atomically-thin semiconductors that could dramatically reduce the electricity demand from AI data centres and high-performance computing.

Scientists to develop atom-thin semiconductors to reduce AI energy costs Credit: Queen Mary University of London

The team will work in partnership with manufacturers and several research institutions to develop new materials and prototype revolutionary low-energy-consumption electronic devices such as transistors.

Sir Colin Humphreys, Professor of Materials Science at Queen Mary University of London, who leads the project, said, “Governments around the world are spending billions building wind, solar, nuclear and gas power stations to meet the huge energy demands of AI data centres. Our approach is to tackle the problem at the source: by reducing the power these centres consume in the first place.

“To do this we will use the latest new materials, called two-dimensional materials, which are atomically thin. This will save over 90% of the energy required by data centres and computers, reduce the cost of electricity, and help to enable Net Zero.”

AI’s energy demands are growing at breakneck speed, and the National Grid predicts that the electricity demand from UK data centres will increase sixfold by 2034, to 30% of total electricity used.

2D semiconductors have emerged as a front-runner technology for building a more sustainable AI industry while also boosting the UK economy.

According to Humphreys, “Leading semiconductor companies including TSMC, Intel and Samsung have already recognised 2D materials as the future, placing them on their technology roadmaps for 2040. Our vision is to make the UK the world leader in ultra-low energy 2D devices well before 2040.”

These 2D materials include graphene and related compounds and can carry electrical charge with far greater efficiency than silicon when scaled down to 2 dimensions, a future requirement for this technology. Electrons in these new materials can move much faster than in silicon, enabling ultra-low power computing and reducing heat waste. They are also suitable for miniaturisation, 3D stacking and new computing architectures like quantum and neuromorphic systems.

“2D semiconductors behave in a fundamentally different way from their bulk (3D) counterparts and their unique electronic properties can support entirely new effects at the atomic scale. We will advance the precise engineering of 2D semiconductors, pushing the limits of what we can create, probe and exploit,” said Amalia Patanè, Professor of Physics at the University of Nottingham.

Humphreys has demonstrated the potential of graphene at industrial scale through his company Paragraf and this project builds on those successes, going from basic science through to prototype devices and enabling the manufacture of complex semiconductor devices made from new 2D materials.