A new range of IoT use cases could be liberated by deployment of stripped-down neural networks in either new systems-on-chip (SoCs), or existing ones already in the field for services such as environmental monitoring. The field of Tiny AI, or Tiny machine learning (ML), has been brewing for two years but has reached a critical point of its fermentation as dedicated SoCs come onstream ready with AI capability embedded. These will still be resource-constrained and low cost, yet capable of being configured for a range of autonomous tasks without requiring the previous high level of computer programming or mathematics expertise. At the same time, there will still be demand for upgrading of existing SoCs for AI, even though that involves…