Every time you ask a question to an AI chatbot or scroll through personalized recommendations, energy is spent storing and processing that data. As artificial intelligence spreads deeper into daily life, the world is generating information at an unprecedented rate — and the electricity needed to handle all that data is soaring. Data centers already guzzle enormous amounts of power, and without major breakthroughs, that demand could keep climbing for decades.
Now, researchers at the University of Edinburgh have proposed a solution that could slash the energy required to store digital information. Dr. Elton Santos and his team have developed a new mathematical framework that designs ultrafast magnetic-field pulses capable of switching the tiny magnetic states inside memory chips using far less energy than current technology allows.
The team, based at the Institute for Condensed Matter Physics and Complex Systems, borrowed a tool from mathematics called optimal control theory. Think of it like finding the smoothest, most efficient route between two points — except instead of navigating roads, the math maps out exactly how a magnetic field should change over time to flip a bit of data using the minimum possible energy.
Computer simulations suggest this approach could reduce switching energies by several orders of magnitude compared to today's leading memory technologies, including DRAM and newer types like STT-MRAM and SOT-MRAM. Even more striking, the predicted energy levels bring magnetic memory much closer to something called the Landauer limit — the fundamental physical floor on how little energy is theoretically needed to process a single bit of information.
The work was published in the journal Advanced Materials. Beyond magnetic fields, the researchers say the same mathematical framework could be adapted to work with electrical currents or even ultrafast laser pulses, opening doors for a wide range of future data storage technologies.
"Every digital operation has an energy cost, and that cost becomes increasingly important as AI and data-intensive technologies continue to expand," Dr. Santos said. "By carefully designing how a magnetic field changes in time, magnetization can be switched far more efficiently than with conventional approaches."
While the findings still need to be tested in real-world experiments, the team has already outlined practical guidelines for building devices that could validate their predictions. If the results hold up, they could eventually help make the growing mountain of digital data less of a burden on the planet's power grid.