China Builds Away From Nvidia Dependency

CNBC reported Monday that Chinese companies across automotive, robotics, and artificial intelligence are actively engineering their operations to reduce dependence on Nvidia chips, even where U.S. export restrictions do not apply.

Robovan Startup Signals a Wider Shift

Autonomous delivery firm Zelostech told CNBC it plans to diversify chip suppliers away from Nvidia over the next one to two years. The company currently uses two Nvidia Orin chipsets per vehicle. Finance and investment director Shi Yunjian cited cost as the primary driver, noting that domestically produced alternatives carry significantly lower price tags. Zelostech claims more than 25,000 vehicles operating across 20-plus countries, dwarfing the fleets of better-known rivals. Alphabet-backed Waymo currently runs fewer than 4,000 vehicles, while Chinese competitors Baidu, WeRide and Pony.ai have yet to reach comparable scale.

Automakers and AI Developers Join the Move

The push extends well beyond one startup. Electric vehicle giant BYD recently unveiled its own driver-assist semiconductors, joining Nio and Xpeng in announcing proprietary chip development. Nio’s CEO William Li told CNBC the company is no longer purchasing chips outright, instead renting compute capacity powered by a mix of processors. A Xpeng vehicle co-developed with Volkswagen already uses Xpeng’s in-house Turing chip, while Volkswagen has separately partnered with Horizon Robotics for driver-assist development in China, bypassing Nvidia entirely.

On the AI side, recent models from MiniMax, Kimi, and DeepSeek’s V4 have all been optimized for Chinese-made hardware, including processors from Huawei and Alibaba‘s T-Head chip unit.

The Background Behind the Break

Nvidia’s driver-assist chips are not currently subject to U.S. export controls targeting advanced AI semiconductors. That makes the shift more striking. Even after Nvidia CEO Jensen Huang accompanied President Donald Trump to Beijing last month, Chinese companies show little appetite for deepening Nvidia reliance. Goldman Sachs analysts wrote in a May report that domestic chip adoption is expected to accelerate meaningfully between 2026 and 2028.

A Narrowing Window for Nvidia

Huawei last week disclosed a new scientific methodology for chip development and indicated the approach would feature in forthcoming products, signaling a renewed push after years of U.S.-imposed constraints. Kevin Xu, founder of hedge fund Interconnected Capital, told CNBC Chinese firms will likely still need some Nvidia supply for the next three to five years. However, he argues Beijing has clear strategic reasons to shrink that window quickly. Domestic chips improve only through real-world deployment, and every Nvidia chip that enters the ecosystem works against that feedback loop.

Read Next: Jensen Huang Joins Trump in Beijing as Nvidia Seeks China Market Access

Similar Posts