Neuromorphic computing aims to use bio-inspired computing paradigms to overcome the limitations of digital computers. For example, the order of magnitude higher power consumption of current devices compared to biological neural systems makes currently used von Neumann architectures increasing unsustainable(Schuller et al. 2015; Christensen et al. 2022). Another motivation is the development of materials that learn from and act directly on physical inputs and outputs (e.g., light, forces, or molecules) rather than symbolic information like in silicon-based computing. For example, perception of biorelevant molecules, light or touch is highly relevant for medical diagnostics, environmental monitoring, and robotics(Stern and Murugan 2023). Another unsolved problem relates to the pollution caused by semiconductor manufacturing. Like other structures created by humans, semiconductors are shaped by top-down processes – which is powerful, but generates waste, depletes finite resources, and creates static products with finite lifetimes. Biological systems are constantly regenerated and dynamically recycled into new parts and shapes. We therefore need new materials for neuromorphic capabilities that not only mimic the energy-efficient computational abilities of the biological inspiration but also use materials based on renewable resources.
