Researchers create clockless FPGA chip for brain-like AI computing

Researchers Eric Oliveira Gomes and Damien Rontani have introduced a scalable neuromorphic architecture that leverages the autonomous, time-continuous evolution of clockless (asynchronous) digital circuits. Implemented on commercially available field-programmable gate arrays (FPGAs), their system creates networks of interacting Boolean spiking neurons with configurable excitatory and inhibitory synaptic weights. This design eliminates the need for a global clock, allowing the chip to naturally emulate the continuous-time dynamics of biological neurons. The team demonstrated a complete processing pipeline for spike-encoded data, achieving competitive performance on an audio classification task while operating at significantly lower power consumption than traditional synchronous digital implementations.

This approach offers a practical alternative to dedicated analog neuromorphic systems, which require custom chip fabrication and are difficult to scale. By using standard reconfigurable FPGAs, the architecture can be rapidly prototyped and adapted for different applications. The clockless design reduces energy waste from clock distribution and idle switching, making it particularly attractive for edge AI and battery-powered devices. The work establishes clockless digital hardware as a viable platform for neuromorphic computing, potentially turning low-cost, off-the-shelf chips into energy-efficient quasi-analog processors for brain-inspired machine learning tasks.