DeepSeek is developing its own AI chip, plans to reduce dependence on Nvidia

China’s leading artificial intelligence (AI) company DeepSeek has begun developing its own AI chip to reduce its dependence on US companies.

According to reports, a rival company to ChatGPT is designing a chip that will be used for the inference stage, the process in which pre-trained AI models generate answers to user questions. The chip is being developed to run existing models more efficiently, rather than training new AI models.

If DeepSeek succeeds in its plan, it will have more control over the hardware needed for its AI models while also reducing its dependence on chips made by US companies, especially Nvidia. At the same time, the company will be largely protected from the potential effects of US export restrictions.

According to reports, DeepSeek has joined the world’s major AI companies that are trying to achieve self-sufficiency in the hardware sector. Last month, OpenAI also announced plans to develop its own custom AI chip.

DeepSeek gained unprecedented global fame when it introduced two highly effective AI models a year ago, which quickly gained popularity around the world and even surprised the technology circles in Silicon Valley and Washington.