Scientists synchronise 105,000 nano-oscillators in computing breakthrough

The network of more than 105,000 magnetic nano-oscillators synchronised within 45 nanoseconds, opening a path towards faster, low-power computing and AI hardware

In a breakthrough that could redefine the future of supercomputing and energy-efficient electronics, a team of scientists and researchers has experimentally demonstrated the world’s largest synchronised network of nanoscale magnetic oscillators, paving the way for a new generation of ultra-fast, low-power computing systems.

 

Scientists from the University of Gothenburg in Sweden, the Indian Institute of Technology (IIT) Bhubaneswar and Tohoku University in Japan have established that more than 105,000 nanoscale spin Hall nano-oscillators, with widths of 10–20 nm, can spontaneously synchronise within just 45 nanoseconds. They proved that such ultra-large spintronic networks can operate coherently and scale efficiently for future unconventional computing.

 

The landmark research, published recently in the prestigious journal Nature Nanotechnology, overcomes one of the biggest challenges in unconventional computing by proving that massive networks of tiny magnetic devices can work together in perfect synchronisation within billionths of a second — a feat scientists believe could eventually transform how computers process information.

 

“Unlike conventional semiconductor chips that process data sequentially using billions of transistors, these oscillators interact through spin waves and naturally synchronise with one another. This collective behaviour allows them to perform complex computational tasks at extremely high speeds while consuming substantially less energy,” said Nilamani Behera, assistant professor in the Department of Physics at IIT Bhubaneswar.

 

Researchers used advanced microwave and optical microscopy techniques to directly observe how thousands of these tiny devices spontaneously organised themselves into a single synchronised state. The synchronised network is nearly 1,000 times larger than previously demonstrated coherent spintronic systems, proving that such technology can be scaled up for practical applications. Demonstrations conducted so far had been limited to arrays of up to 64 oscillators.

 

“This achievement is a giant leap in the development of next-generation computing hardware. Our study confirms that even extremely large networks can coordinate remarkably fast, opening exciting possibilities for future computing systems,” said Behera, one of the lead authors of the study.

 

The researchers employed sophisticated microwave spectroscopy and time-resolved Brillouin light-scattering microscopy, enabling them to directly observe how tens of thousands of nanoscale devices synchronised in real time. Their measurements revealed that synchronisation time increased only marginally even as the network size expanded dramatically — from about 10 nanoseconds for arrays of 100 oscillators to around 45 nanoseconds for networks containing more than 100,000 oscillators.

 

The findings also established several performance records. The synchronised arrays exhibited microwave output power that increased proportionally with the number of oscillators, while the signal linewidth decreased inversely with network size, producing exceptionally stable microwave signals with quality factors exceeding one million. Such highly coherent signals are valuable not only for unconventional computing but also for advanced wireless communication technologies and ultra-fast spectrum analysis.

 

According to Behera, the rapid growth of artificial intelligence is placing unprecedented demands on computing hardware, making conventional silicon-based technologies increasingly constrained by power consumption and heat generation.

 

“The demand for computing power is growing rapidly, especially with the rise of artificial intelligence. Our work demonstrates that very large networks of nanoscale magnetic devices can naturally synchronise in just a few billionths of a second. This opens exciting possibilities for developing future computing technologies that are both faster and far more energy-efficient,” he said.

 

The research further demonstrates that these ultra-large oscillator arrays could serve as physical platforms for emerging computing paradigms such as Ising machines and reservoir computing architectures operating at tens of gigahertz. Ising machines are specialised processors capable of solving highly complex optimisation problems encountered in logistics, financial modelling, industrial planning and scientific simulations, while reservoir computing offers a hardware-efficient approach to processing time-dependent data and AI workloads.

 

Beyond computing, the researchers said, the technology could eventually find applications in intelligent transportation systems, high-speed communication networks, real-time data analytics, scientific modelling and advanced AI accelerators. Although commercial deployment remains years away, the successful scaling of synchronised spintronic networks marks an important milestone in the global search for alternatives to CMOS-based electronics.

 

“The rich non-linear transient response, intrinsic short-term memory and high dimensionality of the SHNO lattice make it a promising medium for physical reservoir computing. The nanosecond-scale dynamics and gigahertz operating frequencies imply an exceptionally large processing bandwidth, while the tunable non-linearity and coupling strength provide natural knobs for task adaptation. Future work could harness these properties by encoding time-dependent inputs in the drive current or magnetic field,” they added.