Chinese artificial intelligence startup DeepSeek is reportedly taking one of its biggest strategic steps since bursting onto the global AI scene. According to a Reuters report, the company has begun developing its own AI chip, a move that could significantly reduce its dependence on hardware supplied by Nvidia and Huawei. If successful, the project would transform DeepSeek from being primarily an AI model developer into a company building both software and the chips needed to power it.
The reported effort reflects a growing global trend among leading AI companies to design their own processors instead of relying entirely on third-party hardware. For China, it also carries broader significance as the country continues pushing domestic technology firms to become less dependent on foreign semiconductor suppliers following tighter US export restrictions.
DeepSeek Reportedly Building Its First AI Chip
According to Reuters, DeepSeek’s chip project has been underway for about a year and remains in the early stages of development. Sources familiar with the matter said the company is designing an AI inference chip, which is used to generate responses after an AI model has already been trained. Unlike training chips, inference processors focus on delivering AI outputs more efficiently while consuming less power and reducing operating costs.
The report states that DeepSeek has quietly expanded its semiconductor ambitions by increasing recruitment of chip-design engineers over recent months. Rather than advertising positions publicly, the company has reportedly hired specialists through private recruitment efforts. Sources also claim DeepSeek has been holding discussions with chip designers, foundries and memory suppliers as it continues developing the project.
Although the company has not officially confirmed the reports, the move would represent a significant change in strategy. DeepSeek has traditionally concentrated on advancing AI models rather than building commercial hardware. Creating its own chip could allow the startup to optimise both software and hardware together, improving performance while reducing long-term infrastructure costs.
Why DeepSeek Wants To Reduce Nvidia Dependence
DeepSeek currently relies on processors from Nvidia and, more recently, Huawei to train and run its AI systems. Its widely discussed R1 reasoning model, which attracted global attention for delivering competitive performance at lower costs, was trained using Nvidia’s H800 chips before US export restrictions tightened. Since then, the company has increasingly adapted its models for Huawei’s Ascend AI processors.
Developing an in-house inference chip would provide DeepSeek with greater control over its computing infrastructure while reducing exposure to ongoing geopolitical restrictions affecting semiconductor supplies. It would also place the company alongside several leading global AI developers pursuing similar strategies.
OpenAI recently introduced its first custom inference chip, developed in partnership with Broadcom, while reports have suggested that Anthropic is also exploring dedicated AI hardware. Instead of relying entirely on external suppliers, AI companies are increasingly seeking processors tailored specifically for their own models and workloads.
The shift reflects changing priorities across the industry. As AI services expand rapidly, inference computing has become one of the fastest-growing areas of demand because every user interaction with a chatbot or AI assistant requires inference rather than model training.
Challenges Still Lie Ahead
Despite the strategic advantages, building a competitive AI processor remains one of the most technically demanding tasks in the semiconductor industry. Designing a modern AI chip requires years of engineering, billions of dollars in investment and access to advanced manufacturing facilities. Those challenges are even greater for Chinese companies because US export controls restrict access to some of the world’s most advanced chip fabrication technologies and high-bandwidth memory components.
Industry analysts also believe DeepSeek’s ambitions may largely remain focused on the domestic Chinese market. Analyst Richard Windsor of Radio Free Mobile told Reuters, “Nvidia is at zero in China and staying there. DeepSeek has almost no chance of selling silicon outside of China unless it gets access to leading edge manufacturing.” His comments suggest that while the project could strengthen China’s AI ecosystem, international expansion may prove far more difficult.
The reported chip initiative comes as DeepSeek is also preparing for another major milestone. Reuters recently reported that the company is seeking approximately $7 billion in its first external funding round, potentially valuing the AI startup between $52 billion and $59 billion. If both the fundraising and semiconductor project move forward successfully, DeepSeek could emerge as one of the most influential AI companies outside the United States, further intensifying global competition in both artificial intelligence and advanced chip development.
