【Text by Observers Network, Wang Yi】"Chinese artificial intelligence (AI) companies are breaking through the global stranglehold of the United States on AI and challenging the American AI that is seen as the gold standard." The U.S. "Wall Street Journal" reported on July 1 that more and more international users are turning to Chinese AI large models instead of American products, from Asia, the Middle East, Africa to Europe, from public universities to multinational banks.

According to insiders, HSBC and Standard Chartered in the UK have started internal testing of the AI large model of DeepSeek, a Chinese company. The world's largest oil producer, Saudi Aramco, has recently deployed DeepSeek into its main data center. Although the White House has banned the use of this model on some U.S. government devices under the pretext of "data security," mainstream U.S. cloud service providers including Amazon, Microsoft, and Google also offer DeepSeek to their customers.

The report points out that just like in many other industries, Chinese companies have begun to attract a large number of customers with their AI large models, which have comparable performance but much lower prices.

This is crucial for the future competition between China and the United States in AI technology. As Brad Smith, president of Microsoft, recently stated at a hearing before the U.S. Congress, "The key factor determining whether the United States or China wins this competition is who's technology is most widely adopted around the world," "Whoever gets there first is hard to replace."

China's practical approach to AI is working

OpenAI's ChatGPT is still the most mainstream AI model globally, with Sensor Tower market research showing it has been downloaded 910 million times, while DeepSeek has 125 million downloads. However, Chinese AI large models are continuously catching up in performance and are rapidly closing the gap with significantly lower prices, narrowing the advantage of U.S. companies.

A study released by Harvard University in early June pointed out that China has an inherent advantage in two key resources for AI development—data and talent—which help it keep pace with the United States.

Global AI Large Model Rankings, The Wall Street Journal Graphic

The "Wall Street Journal" analysis said that currently, U.S. AI companies prioritize major breakthroughs in artificial superintelligence, but the Chinese AI industry focuses more on its practical applications. This pragmatic approach has helped Chinese AI models quickly penetrate the global market.

Chinese top AI companies such as Tencent and Baidu have opened source their large models, allowing users to freely modify and deploy them, greatly stimulating the enthusiasm of global developers and companies. Analysts point out that this strategy is also forcing U.S. companies to explain their "closed-source high price" logic.

Alibaba said that over 100,000 models derived from their open-source Tongyi Qianwen have been developed. Last year, Japanese AI startup Abeja chose Tongyi Qianwen instead of similar products from Google or Meta when developing a custom AI model for the Japanese Ministry of Economy, Trade and Industry.

Oleg Zankov, co-founder of Cypriot AI platform Latenode, said that one out of every five users on the platform chooses the DeepSeek model because its "quality is comparable, but the price is 17 times lower," making it particularly attractive for clients in countries like Chile and Brazil with limited funds and computing power.

The research team at the University of Pretoria in South Africa chose to use this model for a pilot project, considering its open-source nature and the advantage of being able to use it offline to ensure data security.

U.S. restrictions on China backfire

The "Wall Street Journal" pointed out that just a few years ago, the AI industries of the United States and China were closely intertwined. In 2018, U.S. investors accounted for about 30% of the $21.9 billion investment in China's AI industry, and Chinese students also flocked to U.S. universities and Silicon Valley companies. But now everything has changed.

To curb China's AI development, the United States has introduced a series of restrictive measures, trying to prevent Chinese AI companies from accessing U.S. chips, algorithms, and capital, and even contemplating stricter measures. In response to U.S. pressure, China has been accelerating the construction of an AI supply chain independent of the United States with unwavering determination.

U.S. restrictions not only failed to stop the rapid development of China's AI, but also caused huge losses to Western chip and other hardware manufacturers. After the Trump administration halted NVIDIA's sales of its H20 chips to the Chinese market, Jefferies, an investment bank, predicted that this move would cost NVIDIA $10 billion in revenue.

The "Wall Street Journal" warned that if Chinese companies' AI models are adopted worldwide, U.S. companies like Google and Meta may face a loss of market share and revenue.

This year, OpenAI has vigorously expanded its overseas markets, opening offices in Europe and Asia. In an article published on the mainstream news platform Substack on June 25, the company pointed out that Chinese AI upstart Zhipu AI is helping build AI infrastructure in Southeast Asia, the Middle East, and Africa. This made OpenAI, which has similar business, very anxious, and they can only resort to old tactics to hype up the so-called "security risks" of Chinese companies.

Industry insiders said that as the AI systems of China and the United States become increasingly fragmented, AI models will make it easier for users to be trapped in information echo chambers filled with false information and propaganda. Looking further ahead, if the two countries lack cooperation on AI safety issues, the global ability to address potential risks of AI in the future will be severely weakened.

Moreover, the lower the dominant position of U.S. AI companies, the less power the United States will have to set global technology standards. Ritwik Gupta, an AI policy researcher at the University of California, Berkeley, said, "If China still relies on the global AI ecosystem, we can participate in governance," "But if not, China will do things its own way, and we won't see or control it."

This article is an exclusive contribution from Observers Network. Unauthorized reproduction is prohibited.

Original: https://www.toutiao.com/article/7522402113645724199/

Statement: This article represents the views of the author. Please express your opinion below using the 【Top/Down】 buttons.