The neuromorphic platform represents a significant leap forward in computing efficiency

Researchers at the Indian Institute of Science (IISc) have developed a brain-inspired analog computing platform that can store and process data in an astonishing 16,500 conductance states within a molecular movie. Published today in the journal NatureThis advancement represents a huge step forward compared to traditional digital computers in which data storage and processing are limited to only two states.

Such a platform could potentially enable complex AI tasks, such as training large language models (LLMs), to be run on personal devices like laptops and smartphones, bringing us closer to democratizing AI tool development. These developments are currently limited to resource-intensive data centers, due to a lack of energy-efficient hardware. With silicon electronics approaching saturation, designing brain-inspired accelerators that can work alongside silicon chips to deliver faster and more efficient AI is also becoming crucial.

“For over a decade, neuromorphic computing has faced many unsolved challenges,” says Sreetosh Goswami, assistant professor at the Centre for Nanoscience and Engineering (CeNSE), IISc, who led the research team. “With this discovery, we have almost found the perfect system, a rare feat.”

The fundamental operation that underlies most AI algorithms is quite basic: matrix multiplication, a concept taught in high school mathematics. But in digital computers, these calculations consume a lot of energy. The platform developed by the IISc team significantly reduces the time and energy required, making these calculations much faster and easier.

The molecular system at the heart of the platform was designed by Goswami, a visiting professor at CeNSE. When molecules and ions move through a film of material, they create countless unique memory states, many of which were previously inaccessible. Most digital devices can only access two states (high and low conductance), without being able to exploit the infinite number of possible intermediate states.

Using precisely timed voltage pulses, the IISc team found a way to efficiently trace a much larger number of molecular motions and map each of them onto a separate electrical signal, forming a complete “molecular log” of different states.

“This project brought together the precision of electrical engineering and the creativity of chemistry, allowing us to control molecular kinetics very precisely inside an electronic circuit powered by voltage pulses on the order of nanoseconds,” says Goswami.

By exploiting these tiny molecular changes, the team was able to create an extremely precise and efficient neuromorphic accelerator that can store and process data in the same place, much like the human brain. Such accelerators can be seamlessly integrated into silicon circuits to improve their performance and energy efficiency.

One of the main challenges the team faced was characterizing the different conductance states, which proved impossible with existing equipment. The team designed a custom circuit board capable of measuring voltages as small as one millionth of a volt, to identify these individual states with unprecedented precision.

The team also turned this scientific discovery into a technological feat. They were able to recreate NASA’s iconic “Pillars of Creation” image from data from the James Webb Space Telescope, originally created by a supercomputer, using just a desktop computer. They were also able to do it in a fraction of the time and energy required by traditional computers.

The team includes several students and researchers from IISc. Deepak Sharma did the circuit and system design and electrical characterization, Santi Prasad Rath did the synthesis and fabrication, Bidyabhusan Kundu did the mathematical modeling, and Harivignesh S designed a bio-inspired neural response behavior. The team also collaborated with Stanley Williams, professor at Texas A&M University and Damien Thompson, professor at the University of Limerick.

Researchers believe that this breakthrough could be one of India’s biggest advances in AI hardware, putting the country on the map of global technology innovation. Navakanta Bhat, a professor at CeNSE and an expert in silicon electronics, led the design of the circuits and systems for this project.

“What is remarkable is how we have transformed the complex understanding of physics and chemistry into a revolutionary technology for AI hardware,” he says. “In the context of the India Semiconductor Mission, this development could be a game changer, revolutionizing industrial, consumer and strategic applications. The national importance of this research cannot be overstated.”

With the support of the Ministry of Electronics and Information Technology, the IISc team is now focusing on developing a fully indigenous integrated neuromorphic chip.

“This is a completely local effort, from materials to circuits to systems,” Goswami says. “We are well on our way to translating this technology into a system-on-a-chip.”