Other developers moved almost as quickly. On January 19, Shanghai-based Z.ai released a no-charge version of its GLM 4.7 model. Two days later, the firm temporarily halted new subscriptions to its AI coding assistant, citing demand that exceeded available processing capacity. Meanwhile, Hong Kong–listed shares of Baidu climbed to their highest level in nearly three years last week after the release of Ernie 5.0, which the company said surpassed Google’s Gemini-2.5-Pro on internal tests.
U.S.–China technology gap appears to narrow
Demis Hassabis, chief executive of Google DeepMind, told CNBC earlier this month that Chinese AI models could be only “months” behind the latest U.S. offerings. Current claims of parity align with that assessment, though most performance data remains self-reported and has not yet been independently verified.
Industry analysts note that the swift cadence of Chinese launches is partly driven by concerns that additional U.S. export controls on advanced semiconductors could constrain future hardware supplies. Shipping models now allows companies to capture market share and gather user feedback before potential chip shortages make further scaling more difficult.
Open-source approach targets emerging markets
Many Chinese AI developers publish their core model weights under open-source licenses or offer inexpensive access, contrasting with the more closed systems prevalent among major U.S. firms. Advocates argue that lower costs and customization options make Chinese platforms attractive to governments and startups in lower-income countries. Microsoft recently cited estimates indicating that DeepSeek usage in Africa is between two and four times higher than in other regions.
Alex Lu, founder of consultancy LSY, said open sourcing is a strategic move to seed overseas ecosystems. “If developers in multiple countries build their applications on Chinese models, those models become embedded and harder to displace,” he noted.
External studies support the cost advantage. A Stanford University AI Index analysis shows that open-source systems can cut deployment expenses by double-digit percentages compared with proprietary offerings, a difference that can be decisive for small businesses.
Imagem: Internet
Domestic competition focuses on user traffic
Analysts say many Chinese companies are prioritizing user acquisition over breakthrough research. Tencent, operator of the ubiquitous WeChat messaging platform, announced on Sunday that it will distribute 1 billion yuan (approximately $140 million) in digital cash gifts via its Yuanbao chatbot during February’s Lunar New Year holiday. The campaign echoes earlier “red envelope” giveaways that helped WeChat become one of China’s two dominant mobile payment services a decade ago. ByteDance and Baidu are running similar promotions to keep consumers engaged with their AI applications.
Morningstar senior equity analyst Ivan Su argues that integrating AI into existing entertainment, gaming and social ecosystems could generate more immediate value than incremental gains on standardized benchmarks. For example, Tencent can embed conversational tools in popular mobile games or streaming apps to boost in-app spending and advertising revenue.
E-commerce integration aims to offset costs
Earlier in January, Alibaba updated its Qwen AI smartphone app to encourage seamless transitions from chat to shopping, food delivery and payments across the company’s retail platforms, including Taobao. Qwen now claims more than 100 million monthly active users. Consultancy founder Lu said a larger Qwen customer base could drive additional traffic to Taobao, helping to absorb the substantial expenses associated with training and running advanced language models.
Despite divergent strategies, the shared objective among Chinese tech firms is clear: secure large user pools before the market consolidates. Those audiences can then provide training data, subscription fees and cross-selling opportunities that fund further model development.
Potential policy headwinds remain
While the recent wave of product launches suggests momentum, tight U.S. export restrictions on cutting-edge graphics processing units continue to threaten the long-term scaling plans of Chinese AI providers. Industry participants are monitoring whether domestic chipmakers can deliver alternatives with comparable performance. Any sustained shortage could slow inference speeds and raise operating costs, factors that may influence future release schedules.
For now, rapid iteration, aggressive pricing and integration with well-established consumer platforms constitute China’s primary levers for competing in the global AI arena. Market response over the coming months will reveal whether those tactics are sufficient to challenge entrenched U.S. incumbents on both technological capability and worldwide adoption.
Crédito da imagem: Dado Ruvic | Reuters
