While humans and classical computers must perform tensor operations step by step, light can do them all at once. Credit: Photonics group / Aalto University

Researchers have demonstrated single-shot tensor computing at the speed of light, marking a remarkable step toward next-generation AGI hardware powered by optical rather than electronic computation.

Tensor operations are a type of mathematical processing that underpins many modern technologies, especially artificial intelligence, but they go far beyond the basic math most people encounter. A useful comparison is the complex movements involved in rotating, slicing, or reorganizing a Rubik’s cube in several dimensions at once. Humans and traditional computers must break these steps into a sequence, while light can carry out all of them simultaneously.

In AI, tasks ranging from image recognition to language understanding depend heavily on tensor operations. As data volumes continue to grow, however, standard computing hardware such as GPUs is being pushed to its limits in speed, scalability, and energy use.

How Light Becomes a Calculator

Driven by the need for faster and more efficient computing, an international research team led by Dr. Yufeng Zhang of the Photonics Group at Aalto University’s Department of Electronics and Nanoengineering has developed a new way to carry out complex tensor calculations using a single pass of light. This technique enables single-shot tensor computing at the actual speed of light.

“Our method performs the same kinds of operations that today’s GPUs handle, like convolutions and attention layers, but does them all at the speed of light,” says Dr. Zhang. “Instead of relying on electronic circuits, we use the physical properties of light to perform many computations simultaneously.”

The team accomplished this by encoding digital information into the amplitude and phase of light waves, turning numerical values into measurable features of an optical field. As these structured light fields move, interact, and merge, they inherently perform mathematical processes such as matrix and tensor multiplications, which are essential to deep learning. Introducing multiple wavelengths allowed the researchers to expand this method so it could support even more advanced, higher-order tensor operations.

“Imagine you’re a customs officer who must inspect every parcel through multiple machines with different functions and then sort them into the right bins,” Zhang explains. “Normally, you’d process each parcel one by one. Our optical computing method merges all parcels and all machines together — we create multiple ‘optical hooks’ that connect each input to its correct output. With just one operation, one pass of light, all inspections and sorting happen instantly and in parallel.”

Passive, Efficient, and Ready for Integration

Another key advantage of this method is its simplicity. The optical operations occur passively as the light propagates, so no active control or electronic switching is needed during computation.

“This approach can be implemented on almost any optical platform,” says Professor Zhipei Sun, leader of Aalto University’s Photonics Group. ‘In the future, we plan to integrate this computational framework directly onto photonic chips, enabling light-based processors to perform complex AI tasks with extremely low power consumption.’

Ultimately, the goal is to deploy the method on the existing hardware or platforms established by major companies, says Zhang, who conservatively estimates the approach will be integrated to such platforms within 3-5 years.

“This will create a new generation of optical computing systems, significantly accelerating complex AI tasks across a myriad of fields,” he concludes.

Reference: “Direct tensor processing with coherent light” by Yufeng Zhang, Xiaobing Liu, Chenguang Yang, Jinlong Xiang, Hao Yan, Tianjiao Fu, Kaizhi Wang, Yikai Su, Zhipei Sun and Xuhan Guo, 14 November 2025, Nature Photonics. DOI: 10.1038/s41566-025-01799-7

Never miss a breakthrough: Join the SciTechDaily newsletter. Follow us on Google, Discover, and News.