On July 7, 2026, Reuters, citing three sources, reported that Chinese AI startup DeepSeek is working on its own AI chip, intended primarily for inference tasks. The development efforts are still in early stages, with DeepSeek contacting external partners and holding discussions with chip design companies, semiconductor fabricators, and memory manufacturers. The company has selectively hired chip design engineers through discreet channels. According to one source, the chip project began about a year ago, coinciding with DeepSeek gaining international attention with its V3 model. Driven by US export controls and the pursuit of cost control, DeepSeek is freeing itself from dependence on Nvidia.