Analysis: China Has Yet to See Large Independent AI Data Companies

AI Proem author Grace Shao wrote on October 9 that despite China's long-term accumulation of vast data, the country has yet to see large independent AI data companies comparable to US firms such as Surge AI, Scale AI, and Mercor. She attributed this to factors including the frontier model and upstream data market taking shape first in the US, leading players keeping data within closed ecosystems, fragmented data across enterprises and traditional industries, and state-owned enterprises' reluctance to hand data to external vendors.

The article cited the National Data Administration, stating that China generated 52.26 ZB of data in 2025, with only 22.13% of stored data classified as structured data. It also quoted Lin Zhenya, CTO of Benyuan Zhishu, estimating that China has approximately 2,000–3,000 annotation companies, with the top three accounting for about 20%–30% of the market, while the top three in the US are estimated to hold 70%–80% (industry estimates). The report also covered corporate financing and M&A cases: The Information said in September 2025 that UniPat was valued at $1 billion pre-funding, and Humanlaya was valued near $1 billion; Qianbaidu acquired approximately 90.91% of Benyuan Zhishu for RMB 300 million in June 2026; Haitian Ruisheng's 2025 revenue rose 59% year-over-year to RMB 377 million.

[ChainCatcher]