Saia building flash-based inference chip claiming 8x capacity, 4x less power than Jetson

Saia, named by TechCrunch as catching VC attention at Pear VC's PearX demo day in San Francisco, is building a chip running AI inference on-device from flash storage to bypass traditional memory. It claims eight times the capacity while using four times less power than Nvidia's Jetson. Test chips are planned for 2027, mass production for 2028.