Brain-inspired computing gets €30M Dutch boost

The Netherlands takes another step to position itself as a leader in energy-efficient computing, as the Dutch Research Council (NWO) awarded €9 million to a consortium of researchers working on computing systems that mimic the human brain. This initiative addresses the critical energy crisis facing modern AI, which currently relies on power-hungry data centers. By moving away from traditional silicon architectures and toward brain-inspired designs, the project aims to achieve energy-efficiency gains.

The 10X-Factor(y) is a consortium led by Professor Beatriz Noheda at the University of Groningen’s CogniGron research center and includes major technical universities such as Eindhoven University of Technology (TU/e), Delft University of Technology (TU Delft), and the University of Twente, alongside industrial giants and startups like Infineon, Innatera, and Snap Inc.

Why invest in brain-inspired computing?

The primary motivation behind this investment is the power consumption of traditional AI. While modern data centers consume vast amounts of electricity to process large language models, the human brain operates on approximately 20 watts. It achieves this efficiency through asynchronous processing and event-based signaling. Unlike standard computers, which move data constantly between a separate processor and memory unit—a process known as the 'Von Neumann bottleneck'—the brain processes information where it is stored.

10X-Factor(y)'s roadmap

The 10X-Factor(y) project aims to replicate these principles by developing hardware that only activates when necessary. Current AI hardware often struggles to 'power down' unused components, leading to wasted energy. By using spiking neural networks (SNNs) and decentralized architectures, the project aims to achieve systems that are up to 10,000 times more efficient than today’s solutions.

This is a fundamental change in the way machines compute. Achieving these targets would allow complex AI tasks to run on small, battery-powered devices at the 'edge' of the network without needing a constant connection to the cloud.

The NWO grant acts as a catalyst for a total project budget of €30 million, with the remainder provided by private industry partners and research institutions. The project is designed as a 10-year roadmap to transition from laboratory research to real-world industrial applications. It integrates previously fragmented efforts into a unified, industry-driven strategy. By pooling these resources, the Netherlands aims to create a critical mass of expertise that can compete on a global scale. 

The strategic importance of this project extends beyond energy savings to include national and European digital sovereignty. The Neuromorphic Computing NL (NC-NL) coalition, which provides the strategic foundation for this initiative, emphasizes the need for the Netherlands to reduce its dependence on foreign hardware providers. By developing its own neuromorphic chip designs and manufacturing processes, the country can secure its economic strength and technological autonomy.

Practical applications

To prove the viability of brain-inspired computing, the 10X-Factor(y) project will develop seven distinct demonstrators. These prototypes will apply neuromorphic principles to diverse fields, including robotics, health monitoring, and industrial automation.

For instance, Federico Corradi at TU/e leads efforts in neuromorphic edge computing, focusing on how these chips can enhance real-time processing in autonomous systems. In health diagnostics, low-power brain-like chips could enable wearable devices to monitor vital signs and detect anomalies without draining their batteries for hours. In robotics, neuromorphic sensors can process visual data much like a human eye, reacting to movement instantly rather than waiting for a central processor to analyze every frame of video.

These demonstrators are essential for showing industry leaders that neuromorphic hardware is ready for commercial use. The project also addresses the 'responsible innovation' pillar, ensuring that these new technologies are developed with ethical and sustainable frameworks in mind. By testing these systems in real-life scenarios, the consortium can refine the hardware and software co-design platforms needed to support a wide range of future AI applications.

The future of hardware

The technical frontier of this project involves sophisticated hardware innovations such as memristors (electronic components) and thin-film materials. Researchers are exploring ways to integrate memory and processing into a single unit to eliminate the energy-intensive data transfer. Recent breakthroughs in materials science, such as the use of hafnium oxide thin films, have shown that AI energy use can be reduced by 70% at the component level.

The Dutch initiative builds on these global trends by focusing on the entire stack, from advanced materials to system integration. The ultimate goal is a future where AI is pervasive, efficient, and integrated seamlessly into the physical world without the environmental cost of current computing methods.