Tencent, Baidu Reveal Tactics to Navigate U.S. AI Chip Curbs

As the U.S. continues tightening restrictions on the export of advanced semiconductors, Chinese tech giants Tencent and Baidu are unveiling strategies to remain competitive in the global AI race. These include stockpiling GPUs, optimizing software, and relying on in-house semiconductor technologies.
The restrictions, affecting chipmakers like Nvidia and AMD, aim to limit China’s access to powerful graphics processing units (GPUs) used in AI model training and inference. Despite this, Tencent President Martin Lau said the company has a "pretty strong stockpile" of high-end GPUs. Contrary to the U.S. approach of expanding GPU clusters, Tencent has achieved effective model training using smaller-scale clusters, helping preserve its chip inventory for future use.
Tencent is also employing software optimization and developing smaller AI models that require less computing power. The firm is exploring custom-designed chips and China-based semiconductors to support growing AI inference needs without relying entirely on imported GPUs.
Also Read: Ericsson Joins Swedish AI Consortium to Build Nation’s Largest AI Supercomputer
Baidu Bets on Full-Stack Strategy and Domestic Innovation
Baidu, China's leading search engine, has highlighted its "full-stack" AI capability—a combination of in-house cloud infrastructure, AI models, and applications, including its ERNIE chatbot. Dou Shen, president of Baidu’s AI cloud business, noted that even without cutting-edge chips, the company’s control over the entire stack enables cost reductions and performance optimization.
Baidu is leveraging its expertise in managing large-scale GPU clusters efficiently and embracing domestically developed AI chips, which Shen said will form the backbone of long-term innovation in China’s AI ecosystem.
China’s Homegrown Chip Push Gains Momentum
While China still trails the U.S. in AI chip technology, it has made notable advances in building a self-sufficient semiconductor ecosystem. According to Gartner analyst Gaurav Gupta, China has shown “decent success” in developing chips, materials, and packaging, despite global limitations. Nvidia CEO Jensen Huang recently criticized U.S. export curbs as a “failure,” arguing that they harm U.S. companies more than they benefit China.