A skyrmion is a nanoscale magnetic structure—a kind of stable vortex of spins—that can be used to store and process information. Unlike traditional electronic signals, skyrmions require very little energy to create, move, or manipulate, making them ideal for building ultra-efficient computing hardware.

In the SkyANN project, skyrmions are used to emulate the behavior of biological synapses and neurons. They serve as tiny, physical information carriers that can perform computations directly within the material, enabling a new form of neuromorphic hardware that is faster and far more energy-efficient than current AI chips.

Unlike digital AI systems based solely on CMOS technology, SkyANN combines skyrmionics and magneto-electric materials with CMOS interfaces. This hybrid architecture mimics the brain's structure more closely and operates with up to 10,000× less energy, while doubling the bandwidth for the same chip footprint.

SkyANN aims to design, prototype, and validate the world’s first deep skyrmionic artificial neural network. This includes building ultra-low power synapses, neurons, and interconnects, all based on skyrmionic physics, and integrating them into a functional hardware platform for edge AI.

SkyANN is a Horizon Europe-funded project led by a multidisciplinary consortium of universities, research centers, and industrial partners. The team brings expertise in device physics, semiconductor engineering, circuit design, and artificial intelligence.

By drastically reducing the energy required to run AI systems, SkyANN contributes to energy-efficient computing and low-carbon semiconductor innovation—both key goals of the EU Green Deal and the European Chips Act.