China’s DeepSeek Developing In-House AI Chip to Cut Nvidia Reliance, Reuters Reports
Summary
- DeepSeek has begun developing an in-house AI chip as it seeks to reduce its reliance on Nvidia.
- DeepSeek has been in talks for about a year with chip design, foundry and memory companies, and is also expanding hiring for semiconductor design engineers.
- Reuters said the success of in-house AI chip development remains uncertain because it requires significant time and money, while U.S. semiconductor export restrictions on China are making access to HBM and other key technologies more difficult.
Forecast Trend Report by Period
Chinese artificial intelligence startup DeepSeek has started developing its own AI chip, Reuters reported.
Citing sources on July 7, Reuters said DeepSeek is building an in-house chip designed for inference, the stage in which AI models generate responses. One source told the news agency the company has spent about a year in discussions with external partners, including chip design, foundry and memory firms.
DeepSeek has also been expanding its hiring of semiconductor design engineers, according to the report. The hiring is being conducted privately rather than through public job postings.
The company has so far used AI chips from Nvidia and Huawei to train its models. If the effort succeeds, DeepSeek could reduce its dependence on chips from both companies.
Reuters said the project still faces uncertainty because developing an in-house AI chip requires substantial time and money. U.S. semiconductor export restrictions on China have also made it harder for Chinese companies to access cutting-edge foundries and high-bandwidth memory, or HBM.
Uk Jin
wook9629@bloomingbit.ioH3LLO, World! I am Uk Jin.