DeepSeek developing its own artificial intelligence chip — Reuters

Lead

Chinese AI startup DeepSeek is developing its own artificial intelligence chip for inference, according to a Reuters report cited by the South China Morning Post and TASS. The chip would be used to generate responses to user queries using a pre-trained model, a process known as inference.

The Hangzhou-based company previously relied on chips from American Nvidia and Chinese Huawei to train and operate its models, as reported by Reuters. Work on its own chip began roughly a year ago, and the project remains at an early stage, with DeepSeek in discussions with external partners.

Coverage Comparison

The South China Morning Post framed the development within a broader industry trend, noting that Chinese AI labs are increasingly pursuing proprietary chips, mirroring a global shift toward software-hardware integration. The report cited two people familiar with the situation, who declined to be named, saying DeepSeek has been quietly hiring chip-design talent without posting public job openings. The Post also cited an earlier Reuters report stating that DeepSeek's plans for a customized AI inference chip began about a year ago and remained at an early stage.

TASS, citing Reuters, reported the core fact that DeepSeek is developing its own AI chip for inference. It added that demand for inference chips is growing rapidly in the AI sector, and that as AI becomes more widespread, the industry's focus is shifting from training models to operating them, allowing for chips that are cheaper and consume less energy.

Key Claims

• DeepSeek is in the early stages of developing its own AI chip for inference, with work beginning about a year ago, as reported by both the South China Morning Post and TASS.
• The company previously used chips from Nvidia and Huawei to train and operate its models, according to Reuters via TASS.
• DeepSeek has been hiring chip-design talent discreetly, without public job postings, according to two unnamed sources cited by the South China Morning Post.
• Arisa Liu, chief director and research fellow at Taiwan Industry Economics Services, said the core motivation for in-house chips lies in pursuing greater hardware-software synergy and lowering long-term operating costs, as quoted by the South China Morning Post.
• Paul Triolo, a partner and technology policy lead at DGA-Albright Stonebridge Group, wrote in his newsletter AIStackDecrypted that proprietary efforts underscored how China's leading model developers increasingly viewed silicon as a strategic extension of the model stack rather than simply another infrastructure input, according to the South China Morning Post.
• The South China Morning Post reported that the strategy of pursuing proprietary chips carries risk due to massive upfront investments.
• Zhipu AI, a Beijing-based developer of the GLM-5.2 model, is in early talks with domestic chip-design companies about tailored AI processors amid a sharp increase in daily token usage, as reported by The Information, cited in the South China Morning Post article.