DeepSeek launches V4 AI model optimized for Huawei chips, reducing China's dependence on Nvidia technology

DeepSeek, the Chinese AI startup that shocked the industry last year with a cheaper alternative to Western AI models, released a preview of its new V4 model on Friday designed to run on Huawei's processors instead of Nvidia's chips — signaling China's effort to build its own AI hardware supply chain.

The V4 model's professional version outperforms most open-source competitors on world-knowledge tests (benchmarks that measure how much real-world information an AI retains), ranking second only to Google's closed-source Gemini-Pro-3.1 — meaning it delivers top-tier performance without relying on Western chip suppliers.

For business professionals and developers, this matters because AI tool costs and availability depend heavily on chip access. If DeepSeek's Huawei-optimized model becomes widely adopted, companies in China and regions with US trade restrictions gain viable AI options, while global AI competition shifts: companies that previously relied on Nvidia may face new pricing pressure or geopolitical barriers to the chips they need.