AI highlights Public evidence, checked Sep 2026
Product Our All-In-One RISC-V architecture unifies vector (RVV) and tensor units into a single, coherent AI pipeline . Describes its own product as AI on its own site: "ai" Sep 2026 ↗ Product Our CPUs and AI engines minimize memory stalls so attention layers and KV-caches stay fed, improving tokens-per-second and efficiency on modern LLMs . Names the technique "llms", on its own site Sep 2026 ↗