US-China AI War Heats Up: Who Will Dominate the World?

The competition between the United States and China in artificial intelligence (AI) development has entered a new chapter. While Washington has long been the pioneer in producing the most technically advanced AI models, companies from Beijing are taking the opposite approach: creating AI that is sufficiently capable but far cheaper, more efficient, and easier for the global market to adopt. This strategy is beginning to pay off. Over the past year, Chinese AI models have been increasingly used by companies, governments, and businesses across various countries, from Southeast Asia to the Gulf region. According to a report by The Washington Post, the AI race is no longer solely determined by who has the best technology, but also by who can deliver technology that people actually use. Cherie Shi, Global Business Manager at MiniMax, stated that the measure of AI success is no longer just about achieving the highest technical benchmarks. ‘The frontier of AI is no longer about hitting technical benchmarks, but about how many people in the real world are using our models every day,’ she said. Her statement reflects a strategic shift among Chinese AI firms, moving their focus from chasing the title of the most advanced AI to becoming the most widely used. Although US companies like OpenAI and Anthropic are still considered leaders in technical capability, analysts suggest this advantage may not last. A report published in May by the JPMorganChase Center for Geopolitics noted that the capabilities of leading AI models are expected to converge, making the differentiators quality of implementation, integration with industry needs, and the ability to attract users. The institute assessed that Chinese AI labs are developing models specifically designed to win the adoption race. Meanwhile, the cost of using AI has become a major concern for many companies. After a rush to integrate AI into business operations, firms are now facing surging costs for token usage or data consumption units. This situation has created a significant opportunity for Chinese AI companies offering cheaper usage fees and more computationally efficient models. An anonymous salesperson from a major Chinese AI firm said many companies do not actually need the world’s most advanced AI capabilities. ‘Many companies only need a fraction of the capabilities offered by Anthropic or OpenAI. If we can provide 80 percent of the benefit at a much lower cost, that is enough. We will capture that market share,’ the salesperson said. Gunja Gargeshwari, Chief Revenue Officer at Bright Data, observed a similar trend, noting that the cost efficiency and token usage of Chinese models are truly remarkable, with an increasing number of her clients beginning to experiment with them. The success of Chinese AI firms is evident not only in technology but also in business expansion. Singapore, for instance, selected Alibaba’s Qwen AI model to support a national AI initiative. In Malaysia, a Chinese technology company is involved in building a 20 billion US dollar smart city project, while Saudi Arabia has partnered with ByteDance and Huawei to implement AI in urban infrastructure. One of the most aggressive companies in expansion is Zhipu AI, also known as Z.ai. Li Zixuan, Product Director at Z.ai, likened the US-China AI competition to a comparison of luxury cars, suggesting that while American AI still leads in quality, Chinese AI is already highly competitive, more affordable, and has a greater chance of widespread use. Despite being on a US trade restriction list due to alleged links to the Chinese military, Zhipu continues to expand its international business, opening offices in Singapore and Dubai and securing various government contracts across Asia. OpenAI itself highlighted Zhipu’s aggressive expansion in a blog post last year, stating the company was trying to ‘embed Chinese AI systems and standards in developing countries before competitors from the US or Europe can do so.’ Zhipu’s flagship product, GLM, is now claimed to have over 4 million users in 218 countries. Beyond low prices, many Chinese AI models are available as open source, meaning companies and governments can download the model and run it in their own data centres without relying on cloud services. According to Cherie Shi, this approach appeals to many governments wanting to ensure all data remains under their own control. Such models are also attractive to companies handling sensitive data, like those in the banking and healthcare sectors. Aleksander Mordvinov, founder of a fintech startup in Thailand, said he built his company’s entire AI system using Qwen, citing the risk of data leaks with cloud services. ‘I cannot take that risk,’ he said. Martijn Rasser, Vice President of the Special Competitive Studies Project (SCSP), noted that US policy uncertainty is also beginning to affect user confidence, with many users feeling concerned.