Chinese technology company DeepSeek is reportedly now developing its own artificial intelligence chip amid the technological arms race.
DeepSeek’s chip is being crafted for inference, which refers to the phase of AI in which trained models generate replies to users’ questions, rather than for creating new models, Reuters first reported Tuesday, citing three anonymous sources. The news comes as almost every leading AI company worldwide is reportedly moving to create custom chips as part of a broader effort to gain a competitive advantage through hardware tailored to bolster the performance of their AI models, Bloomberg reported. (RELATED: Microsoft Bans Employees From Using ‘Chinese Propaganda’ Chatbot)
DeepSeek did not respond to the Daily Caller News Foundation’s request for comment.
The move may potentially decrease DeepSeek’s dependence on using processors from California-based technology giant Nvidia and Chinese tech company Huawei.
Over the last month, government officials in China have held discussions with several major technology companies about possibly imposing restrictions on overseas access to the country’s most advanced AI models, including unreleased ones, Reuters reported on Tuesday, citing three anonymous sources.
{ const adDivToHide2 = document.querySelector(“#dailycaller_incontent_2”); if (adDivToHide2 && dc_noads_page) { adDivToHide2.classList.add(“hide-premium”, “hide-free”); observer2.disconnect(); console.log(“Ad div2 found and hidden”); } }); observer2.observe(document.body, { childList: true, subtree: true }); ]]> 
Chinese artificial intelligence startup DeepSeek is reportedly designing a proprietary processor, joining a growing number of AI companies seeking greater control over the hardware powering their technology.
The proposed chip is intended to handle inference—the process through which an already-trained AI model interprets a request and produces an answer—rather than the computationally intensive work of training new models, Reuters reported Tuesday, citing three sources familiar with the project.
Developing an in-house processor could allow DeepSeek to optimize its hardware around its own models while reducing its reliance on chips supplied by outside companies. The startup currently depends in part on processors produced by American semiconductor giant Nvidia and China-based Huawei.
DeepSeek did not respond to a request for comment from the Daily Caller News Foundation.
The reported project comes as custom semiconductor development becomes an increasingly important part of the global AI race. Major technology companies are investing heavily in chips built specifically for their products, hoping to improve performance, control costs and avoid depending entirely on third-party suppliers.
Amazon devices chief Panos Panay said in early July that the company is creating AI processors for future versions of its Echo speakers, Fire TV devices and other products.
Nvidia, meanwhile, introduced a new “superchip” in May as part of its effort to make Windows computers better suited for personal AI agents. The company envisions a generation of machines in which artificial intelligence functions less like a conventional software tool and more like an active digital assistant.
DeepSeek’s reported plans also arrive amid growing technological friction between Washington and Beijing.
Nvidia CEO Jensen Huang said in September 2025 that he was disappointed by reports that Chinese authorities had instructed major domestic technology companies to stop purchasing certain Nvidia AI processors. Huang argued that American technology should remain available internationally, including within China.
Chinese officials have reportedly considered restrictions of their own. Government representatives recently discussed potentially limiting foreign access to China’s most advanced AI models, including models that have not yet been publicly released, Reuters reported Tuesday, again citing three sources.
DeepSeek emerged as a major player in the industry in January 2025 following the introduction of its mobile chatbot and DeepSeek-R1 large language model. R1 rapidly became one of the most downloaded and actively used models on the open-source AI platform Hugging Face, attracting international attention to the previously lesser-known startup.
The company has since secured considerable financial backing. DeepSeek raised more than $7.4 billion during its first funding round, according to a June report from The Information. The deal reportedly placed the company’s valuation above $50 billion.
Its rapid expansion has also prompted scrutiny in the United States.
An April 2025 report from the House Select Committee on the Chinese Communist Party alleged that DeepSeek transfers American user information to infrastructure in China connected with a company designated by the U.S. government as having ties to the Chinese military.
The committee also accused the company’s AI system of adjusting its responses to conform with Chinese Communist Party narratives and censorship requirements.
Researchers and educational institutions have separately warned users to examine DeepSeek’s data collection and privacy policies before submitting sensitive or personally identifying information to the service.
DeepSeek’s pursuit of a custom inference chip reflects how thoroughly the AI contest has expanded beyond software. Success in the field increasingly depends upon access to powerful, specialized processors—and companies capable of designing their own hardware may be better positioned to withstand supply restrictions, improve efficiency and compete in an increasingly divided global technology market.
