The First Data Annotation Stock: AI Training Data Outsourcing Platform
1) Data annotation services: service fees charged by annotation volume or by project; unit prices and order volumes are
Key Fields
FIELD STAMPS📌 Background
Demand for high-quality annotated data for large model training has surged, and the data annotation industry has moved from behind the scenes to the forefront. In 2026, C.banner listed on the Hong Kong stock exchange by acquiring a controlling stake in Benyuan Zhishu, becoming the “first data annotation stock”; after the acquisition, it injected RMB 150 million in additional capital into it for capacity expansion; according to broker forecasts, its AI data business revenue will climb year by year from RMB 400 million in 2026 to RMB 1.779 billion in 2028, while the parent's footwear business had 2025 revenue of RMB 156 million and net profit up 56.34% year on year (per company financial statements).
👤 Target Customers
AI large model companies, autonomous driving companies, embodied AI teams, etc. that need training data; paid for by model developers.
💰 Revenue Streams
1) Data annotation services: service fees charged by annotation volume or by project; unit prices and order volumes are not publicly disclosed; 2) End-to-end data services: collection, cleaning, and annotation are charged as a bundled project; bundle prices are not publicly disclosed; 3) Data integration products: finished datasets are sold by set or by license; set prices and licensing rates are not publicly disclosed; 4) New categories such as embodied AI and multimodal: charged by project (opportunistic; there is no verifiable data on how much revenue this can contribute).
🧮 Cost Structure
Annotation labor costs (crowdsourcing or bases), data collection equipment, quality control and platform R&D, and sales channel expenses.
🛡️ Moat
Capital trust brought by a scarce Hong Kong-listed asset; long-term partnerships with leading AI customers; economies of scale in annotation quality and delivery efficiency.
🔑 Keys to Success
- Lock in key customers to secure stable orders
- Expand into high-value-added data categories (e.g., embodied AI, multimodal)
⚠️ Risks
- Changes in data compliance regulations
- Automated annotation disrupts the labor-based model
🏢 Cases
- C.banner (“the first data annotation stock”, controlling Benyuan Zhishu)
- Benyuan Zhishu (AI data service provider)
📊 SWOT Analysis
Strengths
- First-mover listing brings brand endorsement
- Parent company's controlling stake in Benyuan Zhishu creates data factory capabilities
Weaknesses
- Reliance on labor-intensive annotation limits gross margin
- Weak bargaining power with AI customers
Opportunities
- Embodied AI and multimodal models bring new data demand
- Rising compliance demand for domestic large model training data
Threats
- AI automated annotation tools replace manual labor
- Synthetic data reduces reliance on manual annotation
- Intense competition