CA25118 - European Network for NeuroMorphic Computing (EuroNMC)

AI has advanced rapidly and enabled a broad spectrum of applications. The most capable models driving these applications rely on billions of parameters and large-scale computing resources, leading to increasing energy use. Addressing this challenge requires progress in AI models and development of energy-efficient hardware.

Neuromorphic computing (NMC) offers a promising solution by drawing on principles of biological computation, including parallel, sparse, and event-driven information processing, in-memory computing, and online learning through various forms of plasticity. As an alternative computing paradigm, NMC offers potential advantages not only in AI but also in scientific high-performance workloads and other emerging energy-efficient applications. Translating these mechanisms into hardware requires advances across the entire stack, from architectures and algorithms to devices and materials.

Realizing this vision requires collaboration across disciplines. Neuroscience provides the models of brain function that inspire new architectures; computer science and systems research adapt these ideas into algorithms and architectures; hardware design translates them into practical implementations; and device research explores novel materials and circuits to support them. No single laboratory can cover this range, making networking, interdisciplinary exchange, and training essential. Despite promising initiatives, the European research landscape remains fragmented, with limited integration across disciplines and insufficient cooperation with industry.

The Action European Network for NeuroMorphic Computing (EuroNMC) aims at building a pan-European network that will i) reduce barriers between disciplines, ii) align bio-inspired concepts with suitable hardware and architectures, iii) identify applications and benchmarks, and iv) strengthen Europe’s role in NMC and AI.