As the U.S. tightens restrictions on key semiconductor exports, Chinese tech giants Tencent and Baidu are adapting their strategies to remain competitive in the global artificial intelligence (AI) race. Both companies are exploring ways to optimize chip usage, stockpile semiconductors, and leverage domestic semiconductor technology to continue advancing their AI capabilities despite export curbs on critical GPUs from companies like Nvidia and AMD.
Tencent’s Strategy: Stockpiling and Software Optimization
Tencent, China’s leading social media and gaming company, is focusing on efficient use of its semiconductor resources. Martin Lau, president of Tencent, revealed during an earnings call that the company has a “pretty strong stockpile” of graphics processing units (GPUs), which are essential for training AI models. Lau explained that, contrary to U.S. companies’ belief that AI progress requires expanding GPU clusters, Tencent achieves good results with a smaller group of high-end chips.
“We should have enough high-end chips to continue our training of models for a few more generations going forward,” Lau said. Tencent is also focusing on software optimization to improve efficiency, enabling them to deploy the same amount of GPUs to perform various AI tasks. The company is exploring smaller AI models that require less computing power, as well as using custom-designed chips available in China to meet its needs.
Baidu’s Full-Stack AI Approach
Baidu, China’s largest search engine and AI company, has adopted a different strategy by focusing on its “full-stack” AI capabilities. This approach combines cloud computing infrastructure, AI models, and applications such as the ERNIE chatbot. Baidu’s president of AI cloud, Dou Shen, emphasized that despite not having access to the most advanced chips, Baidu’s proprietary technology stack allows it to build strong applications and deliver meaningful value in AI.
Shen also highlighted the efficiency of Baidu’s software and the company’s ability to optimize GPU usage. With the demand for massive computing power increasing due to the rise of foundation models, the ability to manage large-scale GPU clusters and utilize GPUs effectively has become a competitive advantage for Baidu. The company is also making significant progress in developing self-sufficient, domestically manufactured AI semiconductors, a move that aims to reduce reliance on U.S. chipmakers.
China’s Push for Domestic Semiconductor Development
In response to export restrictions, China has significantly ramped up the development of domestically produced semiconductors. While experts agree that China lags behind the U.S. in the realm of AI chips, progress is being made, particularly in the development of chips designed and manufactured within China. Gaurav Gupta, a semiconductor analyst at Gartner, noted that Chinese companies are stockpiling chips and making advancements in semiconductor technology, though they still face challenges in competing with the top U.S. players.
“China has been surprisingly consistent and ambitious in this goal, and one must admit that they have achieved decent success,” Gupta said. The development of a self-sufficient semiconductor ecosystem in China is becoming an important avenue for securing AI chips, even if they don’t yet match the capabilities of chips from U.S. leaders like Nvidia and AMD.
The Future of U.S.-China Semiconductor Trade
Despite these efforts, U.S. executives, including Nvidia CEO Jensen Huang, have called for an end to export restrictions, arguing that they are hurting American businesses more than China’s progress. Huang described the curbs as a “failure,” stating that they are doing more harm to U.S. companies than to China. The ongoing semiconductor race between the U.S. and China continues to be a focal point of global tech competition, with both nations vying for dominance in the rapidly growing AI sector.
As Tencent and Baidu continue to innovate and adapt in response to U.S. restrictions, China’s efforts to develop its own semiconductor ecosystem may play a crucial role in maintaining its competitive edge in the AI race.
