Chinese startup DeepSeek is advancing its strategic shift by developing a specialized AI chip designed for inference tasks, according to three sources familiar with the project.

This move aims to reduce reliance on Nvidia and Huawei chips, which have been critical for training and running DeepSeek's globally popular models.

The new chip focuses on generating user responses during AI model operation rather than training new models, marking a significant departure from traditional semiconductor development.

If successful, this initiative could solidify DeepSeek's position as China's leading AI company while adding complexity to challenges faced by tech giant Huawei.The project highlights growing competition in the global AI chip market, where countries are increasingly prioritizing self-sufficiency.

Experts note that inference-focused chips are essential for deploying large language models efficiently, making this development a key step in DeepSeek's broader technological ambitions.

The initiative also reflects China's broader strategy to dominate emerging technologies amid geopolitical tensions, potentially reshaping the landscape of AI hardware innovation.