China's DeepSeek announces development of indigenous AI accelerator chip to reduce dependence on Nvidia exports and US export controls.

DeepSeek, one of China's leading AI startups, is developing its own artificial intelligence chip to reduce reliance on US firms, according to reports.

The ChatGPT rival is designing the chip for inference—the stage of AI computing in which a trained model generates responses for users—rather than for training new models. The in-house chip would give DeepSeek greater control over the hardware powering its models, making it immune to export controls on critical hardware built by US firms like Nvidia.

DeepSeek joins other global AI developers seeking hardware independence, with OpenAI unveiling its own custom chip last month. The company rose to global fame over a year ago after releasing two highly efficient AI models that went viral worldwide, surprising many in Silicon Valley and Washington, though it has kept a low profile despite becoming a standard-bearer of China's AI ambitions.

An inference chip would target the fastest-growing segment of AI computing demand. As AI applications spread, the industry's computing work is shifting from training models to running them, which relies on specialized chips that can be cheaper and less power-hungry than general-purpose GPUs. However, designing a competitive AI chip typically takes years and significant capital. Manufacturing poses another hurdle: the US bans Chinese designers from accessing the most advanced overseas foundries, while separate US curbs have cut China's access to high-bandwidth memory—a component critical to AI inference chips.

The Hangzhou-based company has increased hiring of chip-design engineers in recent months, though recruitment has been done privately without public job postings. DeepSeek founder Liang Wenfeng said in a rare 2024 interview with a Chinese media outlet that chip export controls were a challenge for the company.

DeepSeek has used both Nvidia and Huawei chips. The foundation model underpinning R1, the reasoning model whose low-cost performance triggered a rout in US tech stocks in January 2025, was trained on Nvidia's H800—a chip designed for the Chinese market that Washington banned in late 2023.

Shares of US-based Nvidia slipped about 1.6 percent in premarket trading following news of DeepSeek's plans. "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," said analyst Richard Windsor of Radio Free Mobile, adding that the development does not affect the chipmaker.