Recently, Chinese artificial intelligence (AI) companies have been gaining significant traction among businesses and developers, thanks to their low-cost models and open-weight technologies. Some notable AI models making waves in the market include DeepSeek, Alibaba’s Tongyi Qwen, and Moonshot AI’s Kimi. Analysts believe that this trend is putting pressure on the pricing strategies of American AI companies. However, it is widely acknowledged that the leading models from the United States still outperform their Chinese counterparts in terms of performance. Nonetheless, Chinese firms continue to grapple with controversies related to valuation, data security, and technology sourcing.
According to a report from a British publication, a London-based tech entrepreneur, Tim Creswick, has been utilizing a set of tools composed of around six AI chatbots for managing daily tasks. One of these tools is Alibaba’s low-cost AI system, Tongyi Qwen, which operates on his own Nvidia processor.
Creswick noted that by using Tongyi Qwen for straightforward tasks and routine programming, he has reduced his expenses on more powerful models from companies like Anthropic and OpenAI by more than half. He metaphorically described using the most advanced “Frontier Models” to sort emails or take notes as akin to “driving a Ferrari to the supermarket.”
He further expressed that he has no qualms about sending data to China for everyday tasks that do not involve sensitive information or code.
Neil Shah, an analyst at Counterpoint Research, pointed out that the service costs of some Chinese systems could be reduced by 80% to 90%, which is driving down market prices. A report from the independent research organization Artificial Analysis revealed that the cost of completing a high-intensity task with China’s Moonshot AI’s K3.1 Pro model is roughly $1.98, compared to about $3.26 for OpenAI’s Astra and up to $7.63 for the higher-performing Fable model from Anthropic. Additionally, it was reported that certain Chinese models process vast numbers of tokens—basic units for text and information processing—at prices significantly lower than those of America’s top models, sometimes by an order of magnitude or more.
However, it is essential to note that lower costs do not imply equivalent performance. The report emphasized that it is generally accepted that Anthropic’s most powerful technology still surpasses that of China’s best models at present. Kyle Chan, a researcher at Brookings Institution, stated that for most users, Chinese models “are good enough and cheaper,” indicating that the competition is primarily in the domain of everyday applications that do not require top-tier performance.
Chinese enterprises have been early adopters of open-weight tools, which allow users to download and modify the technology at no cost according to their needs. Tech-savvy users can even run these models on their own hardware, eliminating reliance on cloud services, which further reduces costs.
In contrast, leading AI tools in the United States typically operate on systems with secure safeguards, with conditions strictly controlled by the development labs. Each mode has its advantages and trade-offs.
Reports indicate that both Anthropic and OpenAI slashed the costs of utilizing some of their cheaper systems by 40% to 50% in September, largely viewed as a response to the price competition. In its recent filing for public listing, Anthropic did not list Kimi or DeepSeek as potential competitors but acknowledged that open-weight AI could “capture part of the market for specific uses” and “put price pressure on our products.”
Chris Wood, an analyst at Jefferies, suggests that, in the long run, this situation could lead to considerable capital losses in the American market, with some market shares shifting to lower-priced Chinese open-source models.
As a result, U.S. tech stocks have recently been under pressure. On October 30, the Nasdaq index fell nearly 3%, and the day before, Google’s parent company Alphabet saw its market value evaporate by $500 billion. The market is increasingly concerned that the substantial investment plans in the AI industry may not translate into sustainable revenue. Notably, Anthropic alone has committed to investing over $550 billion in data centers and other infrastructure.
Such concerns are not unprecedented. In January 2025, when China’s AI model DeepSeek first unveiled its technology, it sparked a wave of sell-offs in financial markets due to fears that its lower operational costs could undermine the prosperity of the American AI sector. Nvidia’s stock plummeted by 17% in a single day, losing nearly $600 billion in market value.
Since then, the valuations of U.S. AI chip companies and major cloud service providers have seen a recalibration and subsequent rise. Companies like Anthropic and OpenAI have been fiercely competing for leading positions on various AI rankings, with their chatbot capabilities driving widespread adoption of AI among enterprises.
Chinese AI firms are also facing significant pressures:
Concerns over Valuation: Since summer, the market capitalization of China’s key tech stock index, the STAR 50 Index—focusing on AI and semiconductor leaders—has evaporated by approximately $500 billion, representing a decline of around 40%.
Data Trust Issues: Major Western corporations are reluctant to entrust their data to Chinese AI firms.
Controversies Over Distillation: Reports have indicated that certain Chinese AI companies have been found attempting to imitate the capabilities of Anthropic and OpenAI’s systems through extensive questioning, a process known as “distillation.” Anthropic has since cut off such access and prohibited Chinese users from utilizing its services. U.S. Treasury Secretary Scott Bessent accused Chinese companies in September of “industrial-scale” theft of technology from American AI laboratories and threatened to implement sanctions.
As the technology community awaits a valuation of $2 trillion for Claude, the developer from Anthropic set to go public, it is reported that Beijing’s Moonshot AI—the company behind the Kimi chatbot—is preparing for a potential IPO as early as next year, with a valuation that could reach $50 billion.
Additionally, DeepSeek, the Chinese AI research organization founded by hedge fund trader Liang Wenfeng, which owns the namesake chatbot, is also planning for an IPO, with its valuation recently approaching $75 billion.
The market will be keenly watching whether these companies can find a balance between pricing, performance, security, and compliance, and how many users are willing to delegate tasks to Chinese AI systems. The outcome of these dynamics will be pivotal in shaping the future landscape of AI development and deployment.