DeepSeek developing in-house AI chip to reduce reliance on Nvidia and Huawei

Planning to gain greater autonomy from its dependence on Nvidia, DeepSeek is reportedly developing its own AI chips. The Chinese startup has long relied on Nvidia's hardware to train and run its globally popular AI models. The new chip is reportedly being designed for inference—the stage of AI computing in which a trained model generates responses for users.

If successful, DeepSeek's expansion into semiconductor development would mark a major strategic shift for a company widely regarded in China as the country's AI champion. The move could also add to the challenges being faced by Huawei as the race to develop homegrown AI hardware intensifies.

Changes with the inhouse Chip

More from Tech

With the introduction of an in-house chip, DeepSeek would join a growing list of global AI developers seeking greater control over the hardware powering their models while reducing their dependence on Nvidia. OpenAI reportedly unveiled Jalapeno, its first custom inference chip developed with Broadcom, last month, while Anthropic is still weighing its options on whether to build its own AI chips, according to Reuters.

For DeepSeek, the effort carries an added strategic dimension. US export controls prevent Chinese companies from purchasing Nvidia's most advanced AI chips, while Beijing has been urging the country's technology champions to develop domestic alternatives.

Dependence on Huawei

DeepSeek has also been steadily shifting toward Huawei's AI hardware. In April, the company introduced its V4 model optimized for Huawei's Ascend chips, while Huawei confirmed that its processors were also used to train V4-Flash, a lightweight version of the model. The launch also boosted demand for Huawei's Ascend 950 chips among Chinese technology companies.

DeepSeek is now reportedly developing its own AI inference chip to tap into the rapidly growing market for running AI models. As AI adoption accelerates, computing demand is increasingly shifting from training large language models to inference, which relies on more efficient and lower-power processors.

However, building a competitive AI chip remains a formidable challenge. Developing advanced processors requires years of research and significant investment. In addition, US export restrictions continue to limit Chinese companies' access to leading-edge chip manufacturing facilities and high-bandwidth memory, a critical component for AI inference hardware.

The chip development effort also comes as DeepSeek prepares to raise outside funding for the first time. The company is reportedly seeking $7 billion in its maiden funding round, which could value it between $52 billion and $59 billion, marking a major departure from its long-standing strategy of avoiding external investors.

DeepSeek first grabbed global attention in early 2025 with the launch of its low-cost R1 reasoning chatbot, which delivered performance comparable to leading US AI models at a fraction of the cost. The chatbot's debut rattled global markets, triggering a selloff in US technology stocks and prompting then-US President Donald Trump to describe it as a "wake-up call" for American AI companies. Built on the company's V3 foundation model, R1 quickly gained traction in China as well as emerging markets across Southeast Asia and the Middle East, establishing DeepSeek as one of China's leading AI startups.