In a world where artificial intelligence is making lives easier, it comes with the cost of massive energy consumption. But a new Monash Warwick Alliance funded initiative is seeking to change that by exploring ultra-thin materials that could one day enable radically more energy‑efficient electronic designs.
Silicon-based computing technologies, built on mid-20th-century designs, were never intended for the massive power demands of the AI era. Computing today accounts for approaching 10 per cent of global electricity consumption, surpassing the aviation sector in carbon emissions, and it's a figure that will accelerate alongside the rapid expansion of AI data centres.
Meeting this challenge requires fundamentally new quantum materials and electronic device designs that can fulfil the promise of AI with dramatically lower power needs. Researchers from Monash and Warwick are working together to create ultra-thin materials that could enable the sustainable continuation of the IT revolution.
Why thin materials matter
Part of the energy problem is driven by the most routine of computing operations: switching transistors on and off. These tiny components underpin almost every digital action, but as chips have become more compact and powerful, the fundamental limits of silicon make further efficiency gains increasingly difficult.
If you've ever felt your smartphone get hot during a long video call or heard your laptop's fans whirring like it's about to take off, you've seen the result of this: our technology loses energy as heat. The same is true in AI data centres.
Ultra‑thin materials with a width only a few atoms offer a compelling alternative to silicon. When electronic pathways are confined to just a few atomic layers, electrical signals can be controlled with far greater precision. This enables transistors that switch cleanly and with less energy loss.
The research team is investigating this possibility through moiré metamaterials. These structures are formed by stacking two ultra‑thin crystals with a slight twist or stretch, so that their atomic patterns no longer line up perfectly. This small misalignment generates an extended, slowly varying pattern known as a moiré superlattice. Although subtle, this gentle, repeating landscape has a profound effect on both atomic structure and electron behaviour, reshaping how they move and interact.
These effects give rise to entirely new electronic and magnetic properties that don't exist in the individual layers alone, creating powerful opportunities for logic and memory technologies.
Computational modeller Professor Nicholas Hine, co-lead of the project from the University of Warwick's Department of Physics, enjoys the fact that the solution to problems created by AI's energy usage may be enabled by AI and machine learning themselves.
We're utilising machine-learning to accelerate simulations to predict and explain the novel quantum mechanical behaviour found in the most promising candidate materials for low-energy, sustainable electronics.
This allows us to predict and design ultra-thin moiré materials and devices with precision before we do anything in a lab.
It creates an exciting closed-loop system, where promising designs move through design, synthesis, characterisation, modelling and experiment phases, and the findings from those feed back into refining our AI models for the next iteration.
Redesigning logic and memory
One of the motivations behind the project is the possibility of rethinking the rigid separation between computation and memory that dominates today's hardware architectures. This is a separation that itself drives energy inefficiency as data is moved between components.
Conventional logic and memory currently rely on entirely different materials, whereas ultra-thin materials host electronic, magnetic, and topological properties that are useful for both next‑generation logic and memory technologies.
This could allow information to be processed and stored within a single material system at extremely low energy cost.
Project co-leader Associate Professor Mark Edmonds from Monash's School of Physics and Astronomy stresses that for now, this vision remains firmly in the realm of foundational research.
The biggest energy challenge in computing isn't in our phones and laptops - it's in the vast data centres that power our digital world. To capture the promise that AI offers, we urgently need hardware that can do more with less.
Our research is laying the foundations for data centres that deliver the same computational power while using only a fraction of the energy, easing pressure on the grid and reducing carbon emissions.
A global solution for a shared challenge
This ambitious work is made possible by the Monash Warwick Alliance, a strategic partnership built on the belief that we are bigger than the sum of our parts and can do more together than alone.
This project also reflects the University of Warwick's broader commitment to responsible and sustainable approaches to artificial intelligence. Warwick is a member of the Quacquarelli Symonds (QS) Responsible AI Consortium, a partnership of 11 world‑leading higher education institutions working together to reimagine how AI can support responsible learning, research, and practice. The consortium provides an important global forum for ensuring that advances in AI‑enabled technologies are matched by ethical, environmental, and societal responsibility.
Professor Michael Scott is the University of Warwick's Pro-Vice-Chancellor (International) and co-Academic Director of the Alliance. He was impressed by the potential to combine Monash's expertise in nanoelectronics with Warwick's world-class materials science capabilities.
The Monash Warwick Alliance was founded on the belief that the world's most pressing challenges require a global response," he explains. "By pooling our infrastructure and intellectual capital, we are driving a sustainable digital revolution, and advancing computing breakthroughs that ensure the AI era doesn't come at the expense of the planet.
Monash University's Pro Vice-Chancellor (Europe), Professor Cecilia Hewlett, is co-Academic Director of the Alliance for Monash. She's conscious of the legacy the project will leave behind, through its focus on training the future leaders in the field. The researchers will establish an international training program for PhD students and early-career researchers that covers the entire pipeline of discovery, from designing new materials to using machine learning for modelling and will teach them to speak each other's scientific languages.
Beyond the science, by funding this project we're developing the people who will lead the quantum revolution," she says. "By supporting joint training programs like this, we are equipping a new generation of researchers with the interdisciplinary skills needed to take these breakthroughs from the lab into the hands of global consumers.
The Alliance funding schemes are designed specifically for this kind of high-impact research. Rather than working in isolation, researchers from Australia and the UK are operating as a single, integrated unit for this project.
Our collaborative research into these exciting materials aligns with the UK Government's AI for Science Strategy and the Australian National AI Plan, both of which prioritise finding sustainable, energy-efficient ways to power the AI revolution.
