Olix

Olix designs silicon for AI inference at the network edge: chips intended to run neural-network workloads on constrained devices, IoT sensors, autonomous vehicles and industrial equipment, rather than in cloud datacentres. Edge inference is architecturally distinct from the in-memory server acceleration Fractile pursues and from the photonic computation Lumai, Optalysys and Salience Labs are building; the AI Hardware Plan named candidates across all three layers of the inference stack rather than backing a single direction.

Olix had been operating for several years before its ministerial endorsement. Edge AI silicon faces a more fragmented customer base than cloud-tier compute, and a government anchor customer across critical national infrastructure or defence applications can compress the sales cycle that typically extends early-stage edge-silicon companies beyond their Runway.