DeepX

When Local AI Couldn’t Run, I Reconsidered Low-Power Inference Chips

DeepX is developing a chip that could run 100-billion-parameter models below 5W, promising for edge devices but still too early to buy for typical ind