Chinese Tech Giants Accelerate Homegrown AI Chip Production Amid Nvidia Uncertainty

Chinese Tech Giants Ramp Up Homegrown AI Chip Production as NVIDIA’s Return RemAIns Uncertain

Chinese companies are accelerating the deployment of domestically developed AI chips, even as reports suggest U.S. chipmaker Nvidia could re-enter the market. On Wednesday, Tencent indicated that production of homegrown chips is set to increase significantly this year, while Alibaba highlighted the expanding use of its self-designed Semiconductors. These developments underscore China's push for self-sufficiency in AI hardware amid ongoing U.S. export restrictions.

Tencent Chief strategy Officer James Mitchell stated that the company expects a "substantial increase" in capital expenditure, particularly in the second half of the year, as more China-designed chips become available on a monthly basis. He noted that the supply of China-developed graphics processing units will "progressively" ramp up throughout the year, supported by growing ouTPUt from manufACTuring facilities within China and neighboring countries.

A number of local chipmakers, including Moore Threads, MetaX, and Huawei, have stepped up efforts to fill the gap left by Nvidia, which has been blocked from selling its most advanced chips to China for over a year. This surge in activity has driven record revenues for Chinese semiconductor firms.

Alibaba is also expanding its footprint in the chip sector. An executive confirmed on an Earnings Call that T-Head, Alibaba's chip design unit, has achieved scaled mass production of proprietary GPU chips. The company framed its in-house chip development as a structural advantage in an environment where access to advanced semiconductors is constrained, supporting both revenue growth and improved gross margins. Alibaba further signaled plans to sell servers equipped with its own chips or co-build computing facilities with other firms, reInforcing its growing role in China's semiconductor landscape.

The comments from Tencent and Alibaba came just before a Reuters report on Thursday claiming that the U.S. has granted APProval for several Chinese firms, including the two tech giants, to purchase Nvidia's H200 chips, one of the most powerful GPUs available. However, the report added that no H200 units have been manufactured yet. It remains unclear whether Nvidia has indeed received U.S. government authorization. When asked about the report, U.S. Treasury Secretary Scott Bessent said the matter was "news to me," adding that such decisions fall under the purview of the Commerce Department.

Over the past year, there have been multiple reports of WashinGTOn permitting Nvidia to ship less powerful chips, such as the H20, to China, though some accounts suggest Beijing has encouRAGed local firms to prioritize domestic alternatives. According to Neil Shah, partner at Counterpoint Research, the push toward "Agentic AI"—AI systems capable of performing more complex tasks—will require more advanced chips, making Nvidia's H200 an attractive option if available. He noted that while China's AI roadmap is pivoting toward domestic-only training infrastructure, the race toward agentic AI has shifted the focus from training to massive Inference Scaling. "Chinese hyperscalers simply cannot afford to wait," Shah said, adding that the moment may be right for Nvidia's H200 to be integrated into a hybrid AI Inference infrastructure combining Chinese and U.S. chips to scale capabilities more rapidly.