For Chinese companies, the question isn’t which AI is the smartest

This report comes from this week’s CNBC The China Connection newsletter, which brings you insights and analysis on what’s driving the world’s second-largest economy. You can subscribe here.

big story

Will the United States or China win the artificial intelligence race? This is a big question for investors wondering where to put their money.

But I often find that many Chinese companies are asking a different question: What AI tools can help me survive in a tough economy?

All of this means that investors who pick AI winners based on smarts alone may be missing the bigger story.

First, there is the cost factor.

Although OpenAI’s GPT-USA Open weight models and DeepSeek are the lowest-cost AI models to run – 30 cents per 1 million tokens – Google’s Gemini 3 Pro sells for $4.50, while Anthropic’s top-of-the-line Claude Opus 4.5 sells for $10, according to data analyzed manually by AI Model Benchmark.

It’s a pay-per-use structure, and whether an enterprise uses OpenAI’s “closed” AI model or remotely accesses DeepSeek or Alibaba’s open-source model, prices can rise quickly.

Instead, many Chinese companies prefer to control costs by choosing a third option—downloading open source models, Suzhou Movitech CEO James Tong told me. His enterprise software company’s clients include Starbucks and Unilever.

Another major difference with the United States is how Chinese companies use artificial intelligence.

Tong said U.S. companies tend to use AI more for coding, while in China they use the technology more in specific situations. For example, since the second half of 2025, Movitech has used AI agents to help manufacturers improve production processes and assist state-owned enterprises with compliance.

Open source models that run locally are again more suitable because they do not require users to upload information to a third-party cloud server.

This is an additional productivity tool when domestic demand is low.

Tong said he had to lower product prices due to economic pressure, but artificial intelligence has improved internal efficiency. Without those gains, the business’s position would be significantly weakened, he said.

Intelligence is not that important

Industry rankings such as manual analysis show that the gold standard for generative AI models still belongs to the United States, but these tools have not yet been officially launched in China.

Beijing’s Great Firewall keeps out Google and Facebook and their AI models, Anthropic cited national security concerns in ban Chinese entity Coming from using Claude, who is widely considered to be the best coding AI out there.

However, discussion of leading American models is widespread on Chinese social media, suggesting such technology is being used by local developers. My conversations with various businesses confirm that they are trying different AI models, including those in the United States, to find one that suits their needs.

“When choosing artificial intelligence solutions, Chinese companies no longer prioritize ‘who is the smartest,'” said Sun Xin, research vice president at Gartner. “Instead, they are looking for reliability, control over illusions, and integration capabilities.”

Productivity improvements are particularly important.

Zhang Wenhao, CEO of Doodod Technology, a Beijing-based consumer marketing consulting firm, said that with the help of artificial intelligence, 10 people can complete the same workload of 20 to 30 people.

As a result, the company’s customer count is roughly equivalent to last year’s more than 100 customers.

Growing global interest

Low prices and easy access also mean that Chinese-made models are gaining more overseas users, especially those who are eager to try them. Viral open source AI agent OpenClaw.

The Chinese model also ranks highly on OpenRouter, a popular platform that allows enterprises to access multiple AI models through a single portal.

While Anthropic and Google’s US models are still The top two ranked by usage last week, One from DeepSeek ranked third, while Kimi’s K2.5, also from China’s Moonshot AI, ranked fourth — only few days after launch.

Moonshot said that since the release of the latest K2.5, its overseas revenue has exceeded the domestic market, and new overseas users have quadrupled.

CNBC reported last month that Moonshot’s valuation reflects how investors want a piece of the pie soared by at least $500 million This comes just weeks after rivals Chipu and Minimax went public in Hong Kong.

As the Chinese New Year approaches, Chinese businesses and technology developers anticipate that DeepSeek may release another groundbreaking artificial intelligence update. Just before the holidays last year, the Chinese company released an open-source model that cost far less to use than OpenAI’s ChatGPT.

DeepSeek’s release at the time Shocked global investorshe believes that without high-end Nvidia chips, China will not be able to innovate in artificial intelligence. Local companies previously excluded from leading U.S. AI models are beginning to experiment with DeepSeek and similar open-source models developed by Alibaba, ByteDance and local startups.

“As long as China’s private sector continues to have room to make money and China’s innovation can keep up with the United States, then people will be interested in Chinese stocks,” said Ren Liqian, director of Modern Alpha at WisdomTree, a U.S. fund management company.

As China’s AI models try to catch up to U.S. rivals, innovation isn’t just about getting smarter, it’s about reaching a price point that makes it commercially viable.

CNBC Hot TV Picks

Carnegie China: Many countries will seek to rebalance relations with China

Ma Daming, director of the Carnegie China Center, said that we now live in an “interest-based world” rather than a values-based world, and he doubts whether there will be a new “golden age” in the economic relations between Britain and China and the United States and China.

China's abandonment of 'three red lines' policy may not be enough to revive real estate market

Zhang Zhiwei, president and chief economist of Pindian Asset Management, said the market has been anticipating policy changes in the real estate market since the fourth quarter of 2025, but questions remain as to whether the reported relaxation of developer lending rules will be enough to change the industry’s momentum.

need to know

Quote of the week

China wants to maintain a balance (how many Nvidia chips it allows to import). They’re investing enough to satisfy the needs of larger companies that really need computing power to continue improving AI, but they’re also very concerned about prioritizing domestic chip manufacturing capabilities.

— Ma Daming, Carnegie China, Director

in the market

Chinese stocks were little changed on Wednesday in a mixed session across the region, but software stocks took a hit as investors weighed how artificial intelligence threatens to automate workflows, squeeze pricing and lower barriers to market entry for rivals.

Hong Kong Hang Seng Index The CSI 300 index was little changed as of 12:45 p.m. ET. So far this year, the two indexes are up about 4.6% and 0.7% respectively.

However, the Hang Seng Technology Index fell about 1.8%. The offshore yuan/dollar exchange rate was little changed at 6.933.

Like their peers elsewhere, Chinese software companies also suffered a sell-off. China stocks Kingdee International Software Plunged over 15%, Cloud Professional Tencent down 3.27%. Alibaba fell by more than 1%, while Baidu fell more than 2%.