South Korean AI chip startup DeepX is preparing for a public offering as it positions its low-power neural processing units as a cost-effective alternative to NVIDIA's data center GPU dominance. DeepX's technology targets edge AI — devices such as cameras, robots, and smart appliances that need on-device inference without cloud connectivity. The company's chips run common vision and language models at a fraction of the energy cost of GPU clusters, making them attractive for industrial IoT and embedded applications. The IPO marks the latest in a growing wave of non-US chip companies seeking to challenge NVIDIA's grip as global AI hardware spending accelerates past $200 billion annually.