China Accelerates Domestic AI Chip Push Despite Nvidia’s Possible Return To Its Market

China’s largest technology companies are signaling a major acceleration in domestic semiconductor deployment, indicating that the country’s artificial intelligence ambitions are increasingly being built around homegrown chips even as reports emerge that Nvidia could regain limited access to the Chinese market.

Fresh comments from executives at Tencent and Alibaba show that Beijing’s drive for technological self-sufficiency is no longer merely a policy aspiration. It is rapidly becoming an operational reality inside China’s AI infrastructure ecosystem.

The shift comes after years of escalating U.S. export restrictions aimed at limiting China’s access to advanced semiconductors critical for training and deploying sophisticated AI systems. Washington’s controls effectively forced Chinese firms to rethink their dependence on Nvidia’s graphics processing units, long regarded as the global standard for artificial intelligence workloads.

Now, Chinese technology giants appear to be building a parallel ecosystem capable of reducing that reliance.

Tencent Chief Strategy Officer James Mitchell said the company expects a “substantial increase” in capital expenditure this year, especially in the second half, as more domestically designed chips become available.

Mitchell said supply of Chinese-designed GPUs would “progressively” ramp up throughout the year, adding that manufacturing output was expanding both inside China and in “neighboring countries,” a phrase that likely reflects Beijing’s broader effort to diversify semiconductor supply chains beyond areas vulnerable to U.S. export pressure.

The comments come boldly because Tencent is one of China’s largest consumers of AI computing power through its cloud, gaming, social media, and enterprise businesses. Increased deployment of domestic chips by a company of Tencent’s scale would provide a substantial commercial boost to Chinese semiconductor firms attempting to challenge Nvidia’s dominance.

China has spent years trying to develop alternatives to Western semiconductor technology, but the U.S.-China technology war dramatically accelerated those efforts. Since Washington blocked Nvidia from selling its most advanced AI chips to China, companies including Huawei, Moore Threads, and MetaX have intensified product launches, fundraising, and public listings in an effort to fill the vacuum.

The momentum is now beginning to show in financial performance. Chinese chip firms have reported surging revenues as local demand shifts toward domestic suppliers under both political encouragement and practical necessity.

Alibaba’s earnings call provided further evidence that China’s AI giants are internalizing semiconductor development rather than depending solely on external suppliers. Executives said Alibaba’s semiconductor arm, T-Head, had achieved scaled mass production of proprietary GPU chips used in the company’s cloud-computing infrastructure.

The company framed its chip development programme not merely as a technological achievement but as a strategic advantage in a constrained global semiconductor environment.

“In an environment of compute scarcity, this structural advantage is favorable to our revenue growth and gross margin improvement,” an Alibaba executive said.

That statement reflects a deeper reality in the global AI race, where computing power has become one of the world’s most valuable resources. As demand for AI infrastructure explodes, access to advanced chips increasingly determines which companies and countries can scale large language models, autonomous systems, and next-generation AI applications.

Alibaba also hinted at ambitions extending beyond internal usage. The company suggested it may eventually sell servers equipped with its own chips or co-build data centers with external partners, signaling its intention to become a broader infrastructure provider within China’s AI ecosystem.

The development aligns with Beijing’s long-term industrial policy objectives, which seek to reduce exposure to foreign technology chokepoints in critical sectors such as semiconductors, cloud computing, and AI.

Yet even as China’s domestic ecosystem expands, Nvidia’s technology remains difficult to replace completely.

Reuters reported Thursday that the U.S. government may have authorised several Chinese firms, including Alibaba and Tencent, to purchase Nvidia’s H200 chips, among the company’s most advanced AI processors. The report said no H200 shipments had yet been produced.

The situation remains uncertain. Asked about the report, U.S. Treasury Secretary Scott Bessent said he was unaware of any approval and noted that export licensing decisions fall under the Commerce Department’s authority.

Over the past year, Washington has repeatedly adjusted chip export rules, sometimes permitting sales of downgraded AI processors such as Nvidia’s H20 while blocking more advanced systems. At the same time, Chinese authorities have reportedly encouraged local companies to prioritize domestic alternatives.

Analysts say Nvidia’s technology would still be highly attractive to Chinese firms because the next phase of AI competition increasingly depends not just on training models but on scaling inference, the computational process required to run AI applications at mass scale.

Neil Shah said China’s transition toward so-called “agentic AI,” systems capable of carrying out complex autonomous tasks, will require more advanced semiconductor performance.

“We are seeing the Chinese AI roadmap pivot towards ‘domestic-only’ with AI training infrastructure,” Shah told CNBC.

However, he added that the competitive battle has shifted toward “massive inference scaling,” where Nvidia’s chips still maintain significant advantages.

“Chinese hyperscalers simply cannot afford to wait,” Shah said, arguing that Nvidia’s H200 chips could become part of a hybrid AI infrastructure combining both Chinese and U.S. hardware.

That hybrid scenario increasingly appears to define the current phase of China’s semiconductor strategy. Beijing is aggressively building domestic capabilities while still attempting to retain selective access to the world’s best foreign technology wherever possible.

The result is a dual-track system in which Chinese firms pursue self-sufficiency not necessarily because domestic alternatives are fully equivalent yet, but because geopolitical uncertainty has made reliance on foreign chips strategically risky.

China was historically one of Nvidia’s largest markets, and continued restrictions threaten billions of dollars in lost revenue while creating space for Chinese competitors to mature technologically. But the broader objective extends beyond replacing Nvidia for China. The country is trying to establish an entirely sovereign AI stack spanning chips, cloud infrastructure, operating systems, and AI models, reducing vulnerability to future sanctions or export controls.

The outcome of that effort could shape the global balance of technological power for years to come.