Chinese artificial intelligence startup DeepSeek is developing its own AI chip, according to people familiar with the project, a move that could reduce the company’s dependence on Nvidia and Huawei while intensifying the global race to control the technology powering the next generation of artificial intelligence.
The project, first reported by Reuters, centers on a custom chip designed primarily for AI inference — the stage where trained AI models generate answers for users — rather than the far more computationally intensive process of training new models. DeepSeek declined to comment on the report.
According to the sources, development has been underway for roughly a year. The company has quietly expanded its hiring of semiconductor engineers and held discussions with chip-design firms, contract manufacturers and memory suppliers as it works to build its own hardware.
The news immediately rippled through financial markets. Shares of Nvidia, whose graphics processors dominate the AI industry, slipped in early trading as investors weighed the possibility that another major AI developer could eventually reduce its reliance on the company’s products.
The timing is significant. U.S. export restrictions have sharply limited China’s access to Nvidia’s most advanced AI chips, pushing many Chinese technology companies to accelerate development of domestic alternatives.
DeepSeek gained international attention after releasing its R1 reasoning model, which surprised many in the technology industry with its strong performance at significantly lower costs than many competing systems. The company originally trained its models using Nvidia’s H800 processors, chips specifically designed for the Chinese market before Washington tightened export controls. Since then, it has increasingly relied on Huawei’s Ascend processors while pursuing greater technological independence.
The move also reflects a much broader shift across the AI industry.
Rather than relying entirely on off-the-shelf processors, leading AI companies are increasingly investing in custom silicon tailored specifically to their own software. OpenAI recently unveiled its first custom inference chip developed with Broadcom, while reports indicate Anthropic is evaluating similar efforts. Inside China, technology giants including Alibaba and Baidu have also invested heavily in proprietary AI processors.
The reason is simple: cost.
Every prompt submitted to an AI chatbot requires computing power. Purchasing chips from outside suppliers means paying those suppliers’ margins while competing for increasingly scarce hardware. A processor designed specifically for one company’s models can reduce operating costs, improve efficiency and provide greater control over future product development.
For businesses, the financial stakes are enormous.
Artificial intelligence is rapidly becoming one of the largest capital investment cycles in technology history. Companies are spending hundreds of billions of dollars building data centers, purchasing chips and expanding cloud infrastructure. Even modest reductions in computing costs can translate into billions of dollars in long-term savings.
Success, however, is far from guaranteed.
Designing competitive AI processors requires years of engineering, advanced manufacturing capabilities and substantial financial investment. Access to leading-edge semiconductor fabrication remains one of China’s biggest challenges under current U.S. export restrictions.
Some analysts remain skeptical about the project’s global impact.
Richard Windsor, founder of Radio Free Mobile, argued that without access to the world’s most advanced manufacturing technologies, Chinese-designed chips may struggle to compete internationally, even if they prove successful inside China’s domestic market.
DeepSeek’s hardware ambitions come as the company reportedly prepares to raise outside capital for the first time. According to recent reports, the startup is seeking approximately $7 billion in funding at a valuation between $52 billion and $59 billion, marking a significant shift after years of avoiding external investment.
For Nvidia, the development highlights the long-term consequences of export restrictions.
While the controls were designed to limit China’s access to advanced American technology, they have also encouraged Chinese companies to accelerate investment in domestic semiconductor development. Every successful homegrown AI processor reduces reliance on imported hardware and strengthens China’s own semiconductor ecosystem.
Whether DeepSeek ultimately delivers a competitive chip remains uncertain.
What is clear is that the battle for AI leadership is no longer being fought only through software. Increasingly, it is becoming a contest over who controls the chips, factories, supply chains and infrastructure that power artificial intelligence itself—a competition likely to shape the global technology industry for years to come.
JBizNews Desk | Beijing
© JBizNews.com All Rights Reserved. Reproduction or distribution without written permission is prohibited.
