China wants to become a leader in AI and is ready to spend billions. How to capitalize on it?

A few years ago, the actions of Chinese AI developers were perceived as an attempt to catch up with American competitors. But now it is already a separate center of power: according to the latest data from Stanford University, China ranks first in terms of the citation rate of AI publications. Beijing's official goal is to become the world leader in AI by 2030. Roman Lukyanchikov, analyst at Freedom Broker, has written about how investors can bet on the growth of the Chinese AI sector.

Several factors have played in favor of the rapid development of AI in China. In addition to a strong private technology sector, which remains the main source of innovation, Chinese AI companies actively benefit from government support.

It's not just about direct funding through regional and national funds (these include the RMB 60 billion National AI Investment Fund and the RMB 1 trillion National Venture Fund, which came into existence last spring), as well as banking programs (such as Bank of China's five-year AI funding program of at least RMB 1 trillion).

Administrative assistance is also available to Chinese companies for the mass introduction of AI into the real economy - primarily in industry, transportation, robotics and healthcare. Officials promote the use of AI by law, create simplified legal regimes, etc.

This support is underpinned by the rapid development of the energy infrastructure needed for future AI computing: in 2025, China will introduce 543 GW of new generating capacity, about eight times more than the US over the same period.

The country's huge domestic potential remains an additional advantage: China is the world's largest digital market, with huge data sets, a wide Internet audience and a rapid cycle of technology testing and deployment.

The speed of Chinese AI development in recent years could have been even faster if not for new problems and challenges. First of all, the Chinese AI industry is hampered by limited access to advanced chips and equipment for their production by the US, Japan and the Netherlands, which leads to a lack of computing power. According to Epoch AI, by mid-2025, the US controlled about 75% of the world's computing power. At the same time, China's share was only 15%; back in early 2022, it was 40%.

Additional hurdles are posed by the low popularity of Chinese AI software and the country's authorities' desire to rely on domestic technology, which will slow the development of Chinese models for at least the next few years.

In addition, regulation of China's AI sector is also a concern - requirements for censorship and ethical principles, data protection and algorithms for data processing could lead to a ban on innovative areas of development or additional costs for companies to monitor compliance with such requirements.

Despite the PRC's rapid progress in artificial intelligence, Chinese AI models are still unable to surpass their American counterparts in terms of "absolute intelligence level," so their strategy is focused on other advantages.

The main task of Chinese AI model developers is to convince businesses that their solutions are more profitable compared to their Western counterparts. Therefore, their main feature is price: since the release of the sensational DeepSeek R1 in early 2025, Chinese companies have been emphasizing their efforts to reduce the cost of information input and output, thanks to which, according to the RAND report, by the beginning of 2026, models from the PRC were already working for only 16-25% of the cost of comparable American solutions.

At the same time, many, unlike American ones, are open source, which gives virtually free access to them for users and accelerates distribution among developers and companies with opportunities for fine-tuning for specific needs.

The second notable feature of China's AI segment is the active implementation of AI models in mass services such as payments and shopping, information search or content creation, so that the user can do "everything in one place". For example, Alibaba's Qwen App allows ordering food and booking travel right inside the chat room without switching between apps, while the WeChat-integrated Yuanbao assistant based on Tencent's Hunyuan and DeepSeek's R1/V3 models can not only answer questions, but also create reminders about any tasks, work with information from the Internet and official social media accounts, and support communication and joint activities in groups.

The Chinese AI market is represented by a huge number of players - as of September 2025, the Chinese Academy of Information and Communication Technology (CAICT) reported that there are more than 5.3 thousand AI companies in the PRC (15% of the global number).

Their greatest successes have been in industrial robotics, computer vision, speech recognition and autonomous driving systems. Due to strong competition, leading Chinese companies in the AI sector often focus on one narrow area (e.g., Baidu is the leader in autonomous driving technologies by a wide margin, SenseTime is a recognized leader in image recognition, and iFlytek dominates in voice technologies). But if we talk about the direction of generative AI models, only about a dozen companies dominate there, which can be conditionally divided into three echelons depending on the scale and monetization.

- Technology giants (Alibaba, Tencent, Baidu and ByteDance)

They are doing "everything at once": developing their own AI models, building data centers, developing chips, offering cloud services, and embedding AI into their established businesses. These companies make money not on the model per se, but on providing an ecosystem within which customers can use AI to their advantage (from online advertising and outreach growth on marketplaces to off-the-shelf enterprise collaboration solutions). In doing so, they have their own distribution channels through which they can "push" AI to hundreds of millions of users (e.g., Baidu is the main internet search engine in the PRC, and ByteDance owns a huge audience through Douyin, a short video service, similar to TikTok, for mainland China).

- "AI Tigers."

They include six AI model startups (Zhipu, MiniMax, Moonshot, 01.AI, Baichuan and Stepfun). They are trying to become "China's OpenAI" and are rapidly increasing their capitalization. These companies make money on sales of the models themselves and products based on them (chatbots, agents, developer toolkits). Their products and business models are more flexible than those of the tech giants, but they depend on the infrastructure of Alibaba Cloud, Tencent Cloud and other large companies, which often both invest in them and provide access to facilities for training and launching models. A prime example is Knowledge Atlas, known as Zhipu. It went public in January this year, and since then its shares have risen in value by about 340%. One of its investors is Alibaba.

- DeepSeek

This startup can be singled out in a separate category. The project grew out of a quantitative hedge fund and was originally conceived as a research lab with access to its own computing power. Although DeepSeek earns money from paid access to its models via APIs and corporate contracts, its expenses are essentially covered by the hedge fund, so the company's goal now is not to maximize profits, but to set a new "benchmark" in terms of price/quality ratio for the market and to displace more expensive solutions.

The prospects for AI development in the PRC make Chinese AI companies attractive to investors. However, they should be viewed as a high-risk growth story. Although the Chinese AI sector trades at a noticeable discount to its US counterparts, high competition in the domestic market is forcing its players to sharply increase infrastructure costs and offer users more favorable terms, which negatively affects margins. This is a forced cost in the battle for future market share. Shares of AI companies, both Chinese and foreign, remain volatile due to regulatory risks, geopolitics and rapid technological changes that regularly reshape industry leaders.

Despite the huge number of AI companies in China, the investment universe here is significantly narrower. Firstly, because of the small number of public companies, and secondly, because of restrictions on foreign investors investing in China's high-tech sector.

For those who want to make an aggressive bet on "pure AI," few highly specialized companies are available, although "AI tigers" are already starting to enter the stock market. In addition to Zhipu, MiniMax began trading on a Hong Kong exchange in January.

A more cautious investor, on the other hand, might look to Chinese tech giants Alibaba, Tencent and Baidu, whose core businesses are not directly related to AI and carry their own risks, but provide the stable cash flow that allows these companies to remain competitive in the AI race for a long time.

Competition for leadership in the global artificial intelligence market will only intensify, both between countries and between individual companies. Therefore, an investor's success in this sector will largely depend on how closely he or she follows the development of technology, new products and the changing balance of power in this complex but extremely interesting market.

This article was AI-translated and verified by a human editor