C.banner (01028.HK) to Acquire Controlling Stake in Benyuan Zhishu to Enter AI Data Sector
Source: EQS|
On
Why data? As public data is gradually digested by large models, the critical bottleneck in AI is shifting from “whose model is bigger” to “who can continuously provide harder, more authentic, high-quality data.” Computing power can be purchased, algorithms can be replicated, but this particular task cannot be bought or expedited. In 2025, Meta invested in Scale AI at a valuation of approximately
The scarcity of Benyuan Zhishu lies in its position and profitability. Founded in 2015, it sits upstream in the AI industry chain and is one of the few domestic suppliers that possess comprehensive data service capabilities, including large models, world models, and embodied intelligence. It serves top-tier large model manufacturers, leading internet platforms, and eminent embodied intelligence enterprises, acting as an exclusive supplier in several high-value data categories. Its revenue in 2025 was approximately
The design of this transaction also demonstrates ingenuity: it is funded entirely with the Group’s internal resources. C.banner’s footwear business generates solid cash flow and maintains a net cash position. Proceeds from the new share subscription will be invested into capacity expansion and R&D, while the founding team will retain equity holdings. Upon completion of the transaction, Benyuan Zhishu will retain its independent brand, independent operations, and data segregation. C.banner will not develop models or compete with Benyuan Zhishu’s clients, allowing industry competitors to continue using the same neutral supplier with confidence.
The management of C.banner stated: “We believe that high-quality data is one of the most critical and scarce infrastructures in the era of artificial intelligence. This transaction allows the Group to enter this rapidly growing sector while maintaining a stable foundation in the consumer industry. We look forward to working with Benyuan Zhishu to continuously provide 'data fuel' of high quality to
|