This report is from this week’s CNBC The China Connection newsletter, bringing you news and analysis on what’s driving the world’s second-largest economy. You can subscribe here.
The big story
Will the United States or China win the race for artificial intelligence? This is the big question that investors ask when wondering where to put their money.
But I’ve often found that many companies in China are asking a different question: What AI tools can help me survive in a tough economy?
All of this means that investors who pick an AI winner based on intelligence alone are likely overlooking a bigger story.
First, there is the cost factor.
While OpenAI gpt-oss the open weight model and DeepSeek are the cheapest AI models to operate – 30 cents for 1 million tokens – Google’s Gemini 3 Pro costs $4.50, while Anthropic’s Claude Opus 4.5 costs $10, according to Artificial Analysis, which compares AI models.
It’s a pay-as-you-go structure whose price can quickly inflate, whether companies use a “closed” AI model from OpenAI or remotely access an open source model from DeepSeek or Alibaba.
Instead, many companies in China prefer to control costs by choosing a third option: downloading open source models, James Tong, CEO of Suzhou-based Movitech, told me. His business software company’s clients include Starbucks and Unilever.
Another key difference from the United States is how Chinese companies use AI.
In the United States, companies tend to use AI more for coding, while in China they use the technology more for specific situations, according to Tong. For example, since the second half of 2025, Movitech has been using AI agents to help manufacturers improve their production processes and help public companies comply.
Open source models that run locally are again more suitable because they do not require users to upload information to third-party cloud servers.
This is an additional productivity tool when domestic demand is sluggish.
Tong said he had to lower product prices due to economic pressures, but that AI had improved internal efficiency. Without those gains, he said, the company would be in a much weaker position.
Intelligence matters less
Industry rankings, such as artificial analytics, show that the benchmark for generative AI models still belongs to the United States. But these tools are not officially available in China.
Beijing’s firewall prevents access to Google, Facebook and their AI models, and Anthropic cited national security concerns in its ban. Chinese-Owned Entities thanks to the use of Claude, widely considered the best AI for coding today.
Still, discussion of major U.S. models is widespread enough on Chinese social media to indicate that these technologies are being used by local developers. And my conversations with various companies confirm that they are trying different AI models, including American ones, to find the one that meets their needs.
“When selecting AI solutions, Chinese companies no longer prioritize ‘who is smarter,'” said Sun Xin, research vice president at Gartner. Rather, they seek reliability, control of hallucinations and the ability to integrate.
Productivity gains, in particular, are essential.
With AI, 10 people can do the same amount of work as 20 to 30 people, said Wenhao Zhang, CEO of Beijing-based consumer marketing consultancy Doodod Technology.
As a result, the company has around 100 customers over the last year.
Growing global interest
Low prices and ease of access also mean that models made in China attract more overseas users, especially among those who drop in to try the model. open source viral AI agent OpenClaw.
Chinese models also rank highly on OpenRouter, a popular platform that allows businesses to access multiple AI models through a single portal.
While the American models of Anthropic and Google were still the the top two by usage over the past week, DeepSeek’s came in third, while Kimi’s K2.5 from Moonshot AI, also from China, came in fourth – just a few days after its launch.
Moonshot said that since the release of its latest K2.5 model, its revenue outside China has exceeded that of the domestic market, with new overseas users quadrupling.
Reflecting how much investors want a piece of the pie, CNBC reported last month that Moonshot’s valuation increased by at least $500 million in just a few weeks – after rivals Zhipu and Minimax listed in Hong Kong.
As the Lunar New Year approaches in China, Chinese companies and tech developers are predicting that DeepSeek could release another revolutionary AI update. Just before last year’s holidays, the Chinese company released an open source model with much lower usage costs than OpenAI’s ChatGPT.
DeepSeek release at the time global investors shockedwho thought China couldn’t innovate in AI without high-end Nvidia chips. Local companies, previously excluded from mainstream U.S. AI models, have begun trying DeepSeek and similar open source models developed by Alibaba, ByteDance and local startups.
“As long as China’s private sector continues to have the opportunity to make money, Chinese innovation can keep pace with the United States, then people will be interested in Chinese stocks,” said Liqian Ren, head of Modern Alpha at U.S.-based fund manager WisdomTree.
As Chinese AI models try to catch up with their American rivals, innovation isn’t just about getting smarter, it’s also about reaching a price point that makes them usable for businesses.
Best TV Picks on CNBC
Damien Ma, director of Carnegie China, said we now live in a world “based on a la carte interests” rather than values, and that he doubts there will be a new “golden age” of UK-China and US-China economic relations.
Zhiwei Zhang, president and chief economist of Pinpoint Asset Management, said markets have been expecting policy changes in the real estate market since the fourth quarter of 2025, but questions remain over whether the announced relaxation of borrowing rules for developers is enough to change the dynamics of the sector.
Need to know
Quote of the week
China wants to balance (how many Nvidia chips it lets in). They’ve left enough room to allow big players who actually need computing power to continue improving artificial intelligence, but they’re also very concerned about prioritizing their domestic chip-building capacity.
— Damien Ma, Carnegie China, director
On the markets
Chinese stocks were little changed Wednesday amid mixed trading in the region, but software stocks took a hit as investors considered how AI threatens to automate workflows, reduce prices and lower barriers to market entry for competitors.
that of Hong Kong Hang Seng Index and CSI 300 were little changed as of 12:45 a.m. ET. Since the start of the year, the two indices are up approximately 4.6% and 0.7% respectively.
However, the Hang Seng Technology Index was down about 1.8%. The offshore yuan was little changed at 6.933 against the greenback.
Chinese software companies have seen a sell-off, among their peers in the region. China’s actions Kingdee International Software plunged more than 15%, while major cloud Tencent fell 3.27%. Alibaba lost more than 1%, while Baidu was down more than 2%.
The performance of the Shanghai Composite over the past year.
Future
February 7: Uruguayan President Yamandú Orsi concludes a nearly week-long state visit