Edinburgh researchers slash memory energy use by orders of magnitude with ultrafast magnetic pulses

Researchers at the University of Edinburgh have developed a theoretical framework using optimal control theory to design ultrafast magnetic-field pulses that could cut the energy required to switch magnetic memory by several orders of magnitude. The breakthrough, published in Advanced Materials, addresses the soaring energy demands of AI-driven data centers by moving memory closer to the fundamental Landauer limit.

Bottom line — The framework brings magnetic memory within striking distance of the Landauer limit, the thermodynamic minimum for processing a bit.

Go deeper (6)

• The method applies optimal control theory to calculate the most energy-efficient path for flipping magnetic states, rather than using conventional brute-force pulses, per the University of Edinburgh.
• Computer simulations suggest the approach could reduce switching energy by several orders of magnitude compared to DRAM, STT-MRAM, and emerging SOT-MRAM devices, according to the paper in Advanced Materials.
• In simulations on van der Waals magnets, switching required as little as 0.94 nanojoules versus up to 91.2 nanojoules for conventional field pulses, with potential to reach the femtojoule range, per SciTechDaily.
• The framework also includes practical guidance for device designs and magnetic-field delivery, offering a pathway to experimental testing, the researchers said.
• Lead researcher Dr. Elton Santos noted the mathematics can be adapted to electrical currents and ultrafast laser pulses, extending its potential beyond magnetic memory, per the university.
• The work comes as data center energy consumption rises sharply with AI expansion; without efficiency gains, ICTs could account for a significant share of global electricity use, the authors warned.

Read the reporting