China’s two largest tech companies are spending substantially in AI infrastructure, anticipating that domestically produced chips will alleviate their supply problems. Alibaba Group Holding and Tencent Holdings said during recent earnings calls that they will sharply increase infrastructure spending. The companies are relying on chips made in China by Huawei Technologies and other local manufacturers to replace limited supplies of American semiconductors. Alibaba’s latest quarterly results showed a major shift in focus. For the three months ending in March 2026, the company reported very low profit. Although revenue increased slightly, Alibaba shifted spending from its main businesses to newer areas such as rapid delivery services and technology development. Profit dropped sharply , with non-GAAP net income falling from 29.847 billion yuan to just 86 million yuan. Eddie Wu Yongming said Alibaba may spend more than 380 billion yuan (US$56 billion) on AI data centers over three years. Analysts say the weaker profits reflect heavy investment in future growth rather than business troubles. Alibaba’s AI products generated 8.971 billion yuan in revenue, representing the 11th consecutive quarter of triple-digit percentage growth. The division’s annual recurring revenue reached 35.8 billion yuan, indicating that the AI industry has progressed past the research phase and is now earning significant cash. Tencent takes measured spending approach Tencent is taking a different path, increasing spending more gradually while maintaining stronger profits. The company spent 31.9 billion yuan on capital projects in the first quarter , a 63 percent jump from the prior quarter. James Mitchell, Tencent’s chief strategy officer, promised “a substantial increase” in spending for 2026, especially in the second half of the year when more Chinese-designed chips are expected to become available. Goldman Sachs analysts predict Tencent’s capital spending will hit 165 billion yuan in 2027, more than twice what it spent in 2025. Both companies face tight supplies of AI chips. “I can tell you that today there isn’t a single card on our service that is idle,” Wu said during Alibaba’s earnings call. The company is deploying its own solution through T-Head, an Alibaba subsidiary that developed an AI chip called Zhenwu. More than 100,000 of these chips are now running on Alibaba Cloud’s platform, with over 30 companies in the automotive and self-driving car sectors using them for research and development. U.S. chip approvals create new complications A recent US policy shift has hampered China’s chip strategy. Washington has permitted seven large Chinese technology companies, including Alibaba, Tencent, ByteDance, and JD.com , to purchase sophisticated Nvidia H200 AI chips under a tight licensing scheme. The H200 is Nvidia’s second-most powerful CPU, based on their Hopper design and optimized for training and running massive AI models. The US license allows eligible companies to purchase up to 75,000 chips each. However, there are certain restrictions. The chips cannot be used for military purposes, the equipment must pass through US ports, and there may be an obligation to split 25% of earnings under US agreements. So far, none of the approved chips have actually been delivered. At the same time, Beijing is urging companies to buy Chinese-made technology instead, such as AI chips from Huawei. This leaves Nvidia caught between the interests of the two countries. Before the export limits were introduced, Nvidia controlled more than 95% of China’s high-end chip market. This situation forces China’s top tech companies to choose between using better and faster foreign technology or supporting the government’s goal of becoming independent in technology. The US Department of Commerce has allowed about 10 Chinese companies to buy these products, but deliveries are still being stopped as China keeps pushing to make its own computer chips. Instead of relying on American companies, Chinese businesses are spending more on building AI systems at home. This could help Nvidia regain access to one of the world’s biggest AI markets, while also showing that the AI competition is now driven by politics as much as by technology. If you're reading this, you’re already ahead. Stay there with our newsletter .
