🔥36氪•Freshcollected in 17m
StepFun establishes new AI subsidiary in Shanghai
💡StepFun is expanding its AI application footprint; watch for new enterprise-grade AI solutions.
⚡ 30-Second TL;DR
What Changed
StepFun fully owns the new subsidiary Qiyue Xingchen.
Why It Matters
This expansion indicates StepFun's strategic move to scale its AI application business and deepen its footprint in the Chinese AI market.
What To Do Next
Track StepFun's upcoming product releases to see how they integrate their proprietary models into enterprise system solutions.
Who should care:Developers & AI Engineers
Key Points
- •StepFun fully owns the new subsidiary Qiyue Xingchen.
- •Registered capital is 10 million RMB.
- •Scope includes AI application systems and algorithm development.
🧠 Deep Insight
AI-generated analysis for this event.
🔑 Enhanced Key Takeaways
- •StepFun (Jieyue Xingchen) was founded by former Microsoft global partner and vice president Jiang Daxin, positioning it as a major player in China's 'AI Six Dragons' (AI六小龙).
- •The company focuses on developing multimodal large models, specifically emphasizing long-context processing capabilities and native multimodal understanding.
- •StepFun has previously secured significant funding from high-profile investors including HongShan (Sequoia China), Meituan, and Vision Plus Capital.
- •The establishment of the Shanghai subsidiary aligns with the company's strategy to expand its physical footprint in key Chinese tech hubs beyond its Beijing headquarters.
- •The new entity, Qiyue Xingchen, is likely to serve as a localized hub for commercializing AI solutions and integrating them into enterprise-level industrial applications.
📊 Competitor Analysis▸ Show
| Competitor | Primary Model | Key Focus | Pricing Model |
|---|---|---|---|
| StepFun | Step-2 | Multimodal/Long-context | API-based/Enterprise |
| Moonshot AI | Kimi | Long-context/Consumer | Freemium/API |
| MiniMax | abab | Multimodal/Voice | API/Enterprise |
| 01.AI | Yi | Open-weights/Global | Open-source/API |
🛠️ Technical Deep Dive
- StepFun's core architecture, Step-2, is a multimodal large model capable of processing complex reasoning tasks and long-context inputs.
- The company utilizes a proprietary training infrastructure optimized for high-throughput, low-latency inference required for real-time multimodal interaction.
- Their technical roadmap emphasizes 'native' multimodality, meaning models are trained on integrated data streams rather than stitching together separate vision and language modules.
🔮 Future ImplicationsAI analysis grounded in cited sources
StepFun will prioritize B2B industrial integration over consumer-facing applications in the Shanghai market.
The subsidiary's stated scope of 'AI system integration' suggests a shift toward providing customized enterprise solutions rather than general-purpose chatbot services.
The company will likely increase its headcount of specialized AI engineers in Shanghai by the end of 2026.
Establishing a new legal entity with 10 million RMB capital typically precedes a significant local hiring drive to support regional operations.
⏳ Timeline
2023-07
StepFun (Jieyue Xingchen) is officially founded by Jiang Daxin.
2024-03
StepFun releases the Step-1 multimodal large model and Step-1.5V.
2024-05
StepFun announces the Step-2 multimodal model, featuring enhanced reasoning capabilities.
2026-07
StepFun establishes Qiyue Xingchen Intelligent Technology in Shanghai.
📰
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: 36氪 ↗