China’s overlooked internet army is quietly embedding AI into everyday businesses
Chinese consumer-facing tech platforms are increasingly developing in-house AI foundation models, moving beyond pure-play AI labs to integrate AI directly into their core business ecosystems.
Intelligence analysis by Gemini 2.5 Flash
While global attention often focuses on China's leading AI research labs, a significant trend sees established Chinese internet companies, such as RedNote, quietly building their own advanced AI systems to enhance everyday services and reduce reliance on third-party providers.
Imagine your favorite apps, like the ones you use to share photos or find cool places to visit, are now building their own super-smart computer brains, called AI, right inside them. Instead of buying these brains from other companies, they're making their own to help you even better with things like finding the perfect outfit or planning a trip, making the apps much smarter and more helpful for you every day.
Analysis
The article highlights a significant, yet often overlooked, strategic pivot within China's vast internet sector: the widespread development of in-house artificial intelligence capabilities by established consumer-facing platforms. While much global attention has been directed towards China's "AI tigers" – dedicated AI labs like DeepSeek and Zhipu AI – a parallel and equally impactful movement is seeing companies spanning e-commerce, social media, and travel build their own foundation models. This internalisation of AI development represents a departure from relying solely on third-party AI providers, signaling a deeper commitment to integrating AI directly into their core business operations and user experiences. The implications of this trend are far-reaching, potentially reshaping the competitive landscape and the nature of digital services in China.
RedNote's Dots Studio
A prime example of this trend is RedNote, the lifestyle platform often likened to Instagram, which recently showcased its advanced AI efforts. Its dedicated AI research arm, Dots Studio, unveiled Dots3-Note Preview, an open-weight model boasting an impressive 280 billion parameters. This release is particularly noteworthy because RedNote is primarily known for its content recommendations in fashion, travel, and shopping, rather than cutting-edge AI research. The company's explicit goal, as stated by Dots Studio, is to "build AI that benefits everyone and helps people solve the many problems they encounter in life," indicating a user-centric approach to its AI strategy.
280-billion-parameter
The scale of RedNote's Dots3-Note Preview model, with its 280 billion parameters, underscores the serious investment and technical ambition of these consumer platforms. The company claims that its system has matched or even surpassed models from globally recognized leaders like OpenAI and Anthropic, as well as domestic front-runners such as DeepSeek and Z.ai, in specific benchmark tests. This suggests that these "overlooked" players are not merely adopting existing AI tools but are actively pushing the boundaries of AI development themselves. The commitment to open-weight models also hints at a potential strategy to foster broader adoption and collaboration within the developer community, further accelerating AI integration.
Operational Shift
Industry analysts view RedNote's move as emblematic of a broader operational shift across China's technology sector. Digital platforms are increasingly opting to develop custom AI solutions tailored precisely to their unique business ecosystems, rather than depending on external providers. This strategic decision allows for greater control over AI capabilities, deeper integration with proprietary data, and the ability to create highly specialized applications that directly address their users' specific needs. This internalisation of AI development could lead to more differentiated services, enhanced competitive advantages, and a more robust, self-sufficient AI ecosystem within China's consumer internet landscape, potentially setting new standards for user engagement and operational efficiency.
Key points
- Chinese consumer tech platforms are increasingly developing in-house AI foundation models.
- This trend is distinct from the "AI tigers" (pure-play AI labs) that have garnered global attention.
- RedNote's AI research arm, Dots Studio, released Dots3-Note Preview, a 280-billion-parameter open-weight model.
- RedNote claims its model matches or outperforms leading domestic and international AI models in specific tasks.
- The move signifies a strategic shift for platforms to embed custom AI into their business ecosystems rather than relying on third-party providers.
This trend could lead to highly customized and efficient AI applications deeply integrated into popular consumer services, improving user experience and fostering innovation within China's digital economy. It allows platforms greater control over their AI capabilities, potentially leading to more tailored and effective solutions for their specific user bases.
The proliferation of in-house AI models by numerous platforms could lead to fragmentation, potential data silos, and increased development costs, possibly hindering broader interoperability or concentrating AI power within a few dominant tech giants. This could also create a more complex competitive landscape for smaller third-party AI providers.
