"Driving Growth Through Price Competition" China's AI B2B Offensive Gains Momentum, Yet Weak Consumer Adoption and Structural Risks Continue to Cap Industry Expansion

Low-cost Chinese AI targets corporate demand for cost savings in the B2B market
ChatGPT maintains a dominant position in the B2C segment, while Chinese models continue to struggle for influence
Government controls raise concerns over talent flight, while AI-driven economic spillover effects remain limited

Chinese artificial intelligence (AI) companies are accelerating their push into the business-to-business (B2B) market. Armed with low-cost offerings and the accessibility advantages of open-source and open-weight models, they have begun capturing growing corporate demand for cost reduction. However, experts argue that this trend is unlikely to translate into an immediate leap forward for China's AI industry. Weak performance in the business-to-consumer (B2C) market and deeper structural constraints continue to weigh on long-term growth prospects.

China's Market Penetration Strategy

According to a June 14 report by the South China Morning Post (SCMP), Alex Yao, JPMorgan's Co-Head of Asia-Pacific Technology, Media and Telecommunications (TMT) Research and Head of China Equity Research, stated in an interview that "the AI monetization race among Chinese technology companies has evolved from a showcase of technological achievements into a survival battle centered entirely on delivering measurable business value." B2B services, particularly coding tools, are increasingly serving as a key growth engine for Chinese AI firms.

Chinese AI companies have recently intensified their efforts to penetrate the B2B market through aggressive pricing strategies, while enterprises seeking to lower AI-related expenses have actively adopted their models. Commenting on the trend, Vishal Misra, Vice Dean for Computing and AI at Columbia Engineering, told The Wall Street Journal (WSJ), "Not every company needs a frontier model," adding that "as open-source models continue to improve, the premium traditionally enjoyed by closed-source AI systems will gradually erode." Open-source and open-weight models currently account for the overwhelming majority of China's AI market.

The pricing gap between advanced U.S. AI models and their Chinese counterparts is clearly reflected in market data. DeepSeek's latest lightweight model, V4 Flash, is priced at $0.14 per one million input tokens and $0.28 per one million output tokens. Its flagship V4 Pro model is offered at between $0.435 and $1.74 per one million input tokens and between $0.87 and $3.48 per one million output tokens. Alibaba's Qwen3.5-Flash costs $0.10 per one million input tokens and $0.40 per one million output tokens, while the higher-performance Qwen3-Max is priced at $0.359 and $1.434, respectively. By contrast, OpenAI's GPT-5.5 reportedly charges $5 per one million input tokens and $30 per one million output tokens. Anthropic's Claude Sonnet 4.6 is priced at $3 for input and $15 for output, significantly above Chinese offerings, while its premium Claude Opus 4.8 model costs $5 and $25, respectively.

Fragile Position in the B2B Market

The challenge is that Chinese AI models have yet to establish meaningful influence in the B2C market. According to mobile application and market intelligence firm Sensor Tower, OpenAI's ChatGPT surpassed one billion monthly active users (MAUs) this year, becoming the fastest consumer application in history to reach the milestone. Over the same period, ByteDance's Doubao recorded 106 million MAUs, while DeepSeek attracted just 68 million.

The traffic gap remains substantial. Data from market research firm Similarweb shows that even during the height of the so-called "DeepSeek Shock," ChatGPT's website received approximately 19 times more visits than DeepSeek. Daily active users (DAUs) on mobile applications were roughly 50 times higher.

Strict censorship is also widely viewed as a barrier to broader consumer adoption. Tests conducted by international media outlets including The Guardian and Ars Technica found that DeepSeek frequently avoided questions related to the Tiananmen Square incident, Chinese President Xi Jinping, and Taiwan, or generated responses closely aligned with official Chinese government positions. Research conducted by AI evaluation firm PromptFoo identified systematic restrictions on responses to more than 1,000 politically and socially sensitive questions. Concerns over privacy and data security remain persistent as well. Several major governments previously suspended downloads of the DeepSeek application after raising concerns that user data was being transferred to China.

Service reliability remains another unresolved issue. DeepSeek experienced significant outages and user registration restrictions as global demand surged. Numerous benchmark studies and media tests found that DeepSeek models exhibited severe hallucination issues, generating incorrect information with high confidence. Multiple reports also documented instances in which responses on politically or socially sensitive topics were abruptly altered or terminated. In addition, Chinese AI models continue to lag behind leading U.S. competitors in ecosystem capabilities that matter most to consumers, including user interface design, plugin support, voice functionality, image generation, and workplace integration.

Structural Constraints Tighten the Growth Ceiling

The limitations inherent in China's state-led growth model are also viewed as a major obstacle to the industry's long-term expansion. According to a Bloomberg report published last month, Chinese authorities have recently begun requiring advance approval for overseas travel by individuals involved in advanced AI development or deemed strategically important to national interests. Those affected reportedly include startup founders, researchers, and senior executives throughout the private AI sector.

Historically, China reserved passport retention and travel restrictions primarily for senior executives at state-owned enterprises, Communist Party officials, nuclear scientists, and researchers at major universities. Extending such controls to private-sector employees represents a significant departure from previous practice.

Sources indicate that individuals are selected based not on rank or employer, but on their perceived strategic importance to China's AI development. This suggests that the scope of restrictions could expand substantially in the future. The measures underscore Beijing's view of AI talent as a critical national strategic asset, while simultaneously introducing risks that could undermine Chinese AI firms' ability to attract and retain top talent. As government pressure intensifies, engineers aspiring to work abroad may choose to leave China earlier in their careers, while overseas Chinese AI specialists may delay or forgo returning home.

The broader economic spillover effects generated by AI companies also remain limited. Although the Chinese government continues to target an urban unemployment rate of approximately 5.5%, reports of quiet AI-driven layoffs are increasingly emerging across multiple industries. On June 10, Reuters reported that a major internet company in Hangzhou began reducing contract workers after mandating the use of the open AI agent platform OpenClaw in March. An engineer at Alibaba Cloud also stated that workforce reductions linked to AI adoption have already begun in some divisions and are likely to proceed through attrition and gradual restructuring rather than large-scale layoffs.

Against this backdrop, experts increasingly believe that AI could emerge as a major threat to employment in China. While the industry may create high-value jobs, the gains are unlikely to fully offset broader job displacement across the labor market. Occupations heavily reliant on repetitive tasks—including customer service, content production, administrative work, and data processing—are expected to face the earliest and most significant disruption. In a recent report, Citibank estimated that approximately 70 million jobs in China, equivalent to 9.6% of total employment, are exposed to AI replacement risk. Among workers in their 20s, the figure rises to 13.6%.