Neuromorphic computing

Modern computing faces a critical sustainability challenge. Artificial intelligence (AI) is driving energy demand to levels that already limit the capacity of data centres and challenge the power grid. It also hinders the use of AI at places where power is limited, as health sensors or devices in remote locations. Instead, data is communicated into the cloud, resulting in severe privacy issues.

These issues cannot be tackled with incremental changes but requires different computing paradigms like neuromorphic computing, which takes inspiration from the extreme energy efficiency of brains for information processing. This approach consists of using one or more of the principles used by the brain: use of analogue signals, implementation of asynchronous (event-based) processing, communication and information encoding via spikes and bringing the processor closer to the memory.

Creating a blueprint for future-proof computing

After 70 years of continuous innovation, today’s challenges demand a radical shift in computing. Neuromorphic engineering is leading this revolution, mimicking the brain’s architecture with spiking neural networks and specialized chips that are up to 10,000 times more energy efficient than traditional processors. By combining past expertise with new materials and designs, neuromorphic computing is transforming AI hardware, enabling ultra-efficient, low-power systems that push the boundaries of what machines can achieve.