Neuromorphic Chip
Readiness
TRL 4 · SRL 5.5 · gap -1.5
Gap = technical readiness − societal readiness. Positive: the technology is ready and society is not. Negative: society wants it more than the technology can deliver.
Not scored. Watchlist entries do not enter any gradient.
Processors built to compute the way neural tissue does — memory and processing co-located, events rather than clock cycles, activity only where signal is present.
The efficiency case is strong and specific. Conventional architectures spend most of their energy moving data between memory and processor; neuromorphic designs largely remove that traffic, and for sparse, event-driven, always-on sensing workloads the energy advantage runs to orders of magnitude. That matters for anything battery-powered that has to watch continuously.
Technical readiness 4 with appetite ahead, and the constraint is not the silicon — research chips with over a billion neurons exist and their designers have said publicly they are not close to commercialisation. The constraint is software. The entire machine-learning toolchain assumes conventional hardware, and a chip that requires a different programming model competes against an ecosystem rather than against a processor.
Common misconception: that neuromorphic silicon competes with GPUs on training. It addresses low-power inference at the edge, and against a GPU on a training run it is not in the contest.
What it does
mechanism — not yet written. Published as missing rather than hidden.