105,000 Nano-Oscillators Synchronized in 45ns: The Largest Spintronic Sync Network Scouts New AI Compute
A Nature Nanotechnology study demonstrates the largest nanoscale magnetic-oscillator sync network: 105,000 spin oscillators cohere in 45ns, ~1000x prior scale — scouting hardware for Ising machines and reservoir computing beyond silicon.
45 nanoseconds, 105,000 oscillators
Per IThome citing Business Standard (July 14), a team experimentally demonstrated the world's largest nanoscale magnetic-oscillator sync network, synchronizing 105,000 nano-oscillators in 45 nanoseconds. Verified by the University of Gothenburg, IIT Bhubaneswar and Tohoku University, the work appears in Nature Nanotechnology.
The device: spin-torque nano-oscillators (STNO)
Nano-oscillators generate high-frequency EM/optical waves via nanoscale physics; the most common are spin-torque nano-oscillators (STNOs), using the spin-Hall effect to drive magnetization oscillation and emit microwave signals. Oscillators here are just 10–20 nm wide. The team observed synchronization directly using microwave spectroscopy and time-resolved Brillouin light-scattering microscopy.
Near-"free" scalability
The key finding is scalability: 100 oscillators synced in 10ns, and scaling to 105,000 raised sync time only to ~45ns. Prior public demos topped out at 64 oscillators, so this coherent spintronic system is ~1000x larger, showing that very-large-scale spintronic networks can stay coherent and keep scaling.
Why it matters for AI
Researchers note AI's growth is driving compute demand while amplifying silicon's power and thermal limits. Such synced arrays have nonlinear transient response, short-term memory and high dimensionality, run at tens of GHz, and can serve as hardware for Ising machines and reservoir computing, with temporal inputs encoded via drive current or magnetic field. It represents a "beyond-silicon" frontier path — computing with a physical system's natural dynamics rather than piling on transistors.