Tagged edge AI · show all articles
Two questions decide how small and on-device AI matures by 2035: whether dedicated hardware closes the capability gap with the cloud, and whether on-device personalization makes privacy the default. Four scenarios, each with a falsifier.
A capable small model can be reached by compressing a large one, training a small one on purpose, searching architecture space for it, or routing around density altogether. These are four different engineering problems, not four contestants in one race.
A quantized or pruned model reported to lose less than a point of accuracy has been shown to average well across a benchmark. Whether it kept the one capability a deployment actually needs is a separate question the average is built to hide.
Before a phone could run a language model, it had to run anything at all: a narrower convolution, a dedicated chip, an always-listening detector, and a trick for compressing what a bigger model already knew. This is that decade, in order.
While most of this cohort tracks gigawatt-scale datacenter deals, SiMa.ai is building AI chips meant to run inside a camera, a robot, or a car — nowhere near a datacenter at all.
Almost every company in this cohort's accelerator and startup tracks is American, Korean, or Chinese. Europe's near-total absence from that list is itself a story — and Axelera AI is one of the few companies trying to close the gap.