Smarter and safer autonomous vehicles.

The current cost of the "full self-drive" feature is about $12K per vehicle. This is because of the current chip capacity and AI training cost. The AI learning approach is highly inefficient and also very limited. The NDPU provides the ability to condense vectors and make the hardware more elegant, thereby requiring much less power and inefficient computations. The NDPU can reduce the need for very expensive hardware as well as AI training costs. Our chips also facilitate our novel approach to AGI. Transforming learning models from a series of in-efficient neural networks to cognitive networks with our holographic architecture approach. This allows automatic bootstrapping and inevitable generalization. Results are 1 cost reduction of hardware cost and less power consumption + hardware usage and 2 increased learning via less training required in new driving terrains, environments and regulations.