Researchers at the University of Tokyo have developed a magnetic switching device that operates at speeds up to 1,000 times faster than current AI accelerators while using significantly less energy and producing minimal heat. This innovation could address overheating and battery drain issues in electronics, paving the way for more efficient computers and smartphones.
The research was published in the journal Science earlier this week and builds on a previous study published in Nature in January 2025. The device demonstrates a method for flipping a binary magnetic state at picosecond speeds, representing a substantial improvement over the nanosecond switching standard in silicon-based processors.
The new device may help solve a longstanding issue with conventional processors: excessive heat generation. As processor speed increases, so does the heat produced, resulting in higher power consumption and stressing data center infrastructure. The study’s authors claim to have addressed this challenge with a novel spintronic device made from a manganese and tin compound (Mn3Sn), an antiferromagnet.
Spintronic devices leverage both the charge and spin of electrons for data processing, storage, and transmission, diverging from traditional semiconductor methods that focus solely on charge. The researchers showcased the device’s capabilities by successfully flipping its magnetic state using a 40-picosecond electrical pulse, which generates significantly less resistive heat compared to conventional computing switches.
Moreover, this device operates with a much lower energy requirement than modern AI accelerators, heightening prospects for developing faster and more efficient AI hardware. One picosecond is one trillionth of a second, which is 1,000 times shorter than a nanosecond.
If the technology can reliably transition from research labs to commercial production, it could find use in cloud-based quantum services, making optical quantum computing more accessible. Professor Tomo Nakatsuji noted, “there is (also) a possibility that data that takes an hour to download can be processed in one second.”
However, the researchers caution that while the binary switching speed has increased significantly, this does not directly correlate to a 1,000 times boost in overall computing speed. The complexity of computer systems means that a faster switch can only contribute to a certain extent in enhancing overall performance.
