Stepfun’s $719M Raise Highlights China’s Shift From AI Models to Real-World Deployment

Backed by state capital and industry players, the startup is positioning itself at the centre of China’s push to embed AI into cars, devices, and physical systems

China-based Stepfun has raised around RMB 5 billion (approximately US$719 million) in a Series B+ funding round, placing it among the most heavily funded private AI startups in the country this year. The company said the capital will be used to advance large-scale foundation model development, roll out AI agents, and support the creation of AI-powered products designed for real-world use.

The startup, which works closely with Geely Automobile Holdings, is part of a growing group of Chinese AI firms shifting focus from model research alone to deployment across physical systems such as vehicles, consumer electronics, and robotics. The round also coincides with Stepfun appointing veteran AI entrepreneur Yin Qi as chairman, a move that signals sharper execution and industry alignment.

The Series B+ financing brought in a diverse group of backers spanning state-linked capital, regional government funds, financial institutions, and industrial investors. Returning investors such as Tencent, 5Y Capital, and Qiming Venture Partners joined the round, alongside new participants including Pudong Venture Capital, China Life Private Equity Investment, and Shanghai State-owned Capital Investment Leading Fund.

The composition of the investor base is notable. Conservative institutional capital and local government funds are increasingly selective in AI investments, suggesting Stepfun is viewed not just as a research-driven startup, but as one capable of commercial deployment and industrial relevance.

Leadership change follows major funding round

Alongside the funding, Stepfun announced the appointment of Yin Qi as chairman. Yin is best known as a co-founder of Megvii, one of China’s early computer-vision unicorns, and currently serves as chairman of Qianli Technology, a Geely-backed intelligent driving company.

The company said Yin will collaborate with founder and chief executive Jiang Daxin, chief scientist Zhang Xiangyu, and chief technology officer Zhu Yibo in setting the firm’s strategic priorities and product roadmap.

“Qianli is building AI-enabled application scenarios centred on vehicles,” Yin said in a recent public discussion. “Stepfun provides the underlying foundation models — the core AI capabilities that make these applications possible.”

Focus shifts toward deployment and commercial use

Founded in Shanghai in 2023 by Jiang, a former Microsoft executive, Stepfun focuses on foundation models designed for deployment across cars, smartphones, robots, and other devices. From early on, it has maintained close ties with Geely and Qianli, jointly developing smart cockpit systems already used in mass-produced Geely vehicles.

At the core of Stepfun’s strategy is its flagship Step 3 large language model, alongside multimodal systems capable of processing text and visual inputs together. The company says these models are designed with high inference efficiency and broad hardware compatibility, allowing them to run both in the cloud and directly on devices.

This emphasis reflects a broader industry trend: as compute costs rise and regulation tightens, AI companies are under pressure to show how models perform outside controlled environments.

Strategy centres on embedding AI into devices

Rather than positioning itself purely as a cloud AI provider, Stepfun is pushing into what it calls an “AI plus terminals” approach — embedding AI directly into hardware.

Key focus areas include:

• Automotive AI, where its end-to-end voice model powers Geely’s Agent OS smart cockpit
• On-device AI, reducing reliance on cloud inference
• Multimodal interaction, linking perception, reasoning, and action

Local media have reported that Stepfun’s models are expected to be deployed in more than one million vehicles this year, underscoring the company’s emphasis on scale and production use.

Deal size stands out amid sector consolidation

The scale of Stepfun’s latest funding comes at a time when China’s artificial intelligence sector is becoming more selective. While several AI developers have turned to public markets or delayed private fundraising, capital in the private market has increasingly flowed toward a smaller group of companies seen as having clearer commercial paths.

Industry data show that the number of AI funding deals in China has declined over the past year, even as average deal sizes have increased. This suggests investors are placing fewer bets, but committing larger sums to companies they believe can move beyond research and into sustained deployment.

In Stepfun’s case, the investor mix reflects a preference for AI firms with established industrial relationships, operational use cases, and management teams experienced in scaling technology businesses. Rather than focusing solely on model performance or technical benchmarks, investors appear to be weighing factors such as integration with existing industries, long-term application potential, and alignment with national and regional priorities.

The round highlights a broader shift in how AI companies are evaluated in China’s current funding environment, where commercial readiness and real-world deployment are increasingly central to investment decisions.

Capital follows deployment

Stepfun’s Series B+ round reflects a broader recalibration underway in China’s artificial intelligence market. After years of heavy investment in model development and technical benchmarks, investors are increasingly prioritising companies that can demonstrate how AI is embedded into real products, regulated industries, and physical systems at scale.

The composition of Stepfun’s investor base — spanning state capital, industrial players, and established technology funds — underscores this shift. It suggests that capital is now being deployed with longer time horizons and clearer expectations around commercialisation, ecosystem alignment, and national industrial goals, particularly in sectors such as automotive, consumer electronics, and advanced manufacturing.

At the same time, the round highlights rising barriers for newer or less connected AI startups. As funding concentrates around a smaller group of perceived leaders, access to compute, data, and large enterprise partners is becoming as important as model performance itself.

Whether Stepfun can translate its funding, partnerships, and leadership into durable advantages remains an open question. But the deal offers a clear signal of how China’s AI market is evolving — away from experimentation, and toward a phase where scale, integration, and execution determine which companies continue to attract capital.

Quick Takeaways

• Stepfun raised about US$719 million in a Series B+ round, making it one of the largest private AI financings in China this year.
• The round attracted a mix of state-backed funds, regional government investors, industrial players, and established technology funds, highlighting growing selectivity in China’s AI capital market.
• Stepfun is shifting focus from model research to real-world deployment, particularly across vehicles, consumer devices, and other physical systems.
• The appointment of Yin Qi as chairman strengthens the company’s leadership as it moves toward commercial execution and deeper industry integration.
• Investor interest reflects a broader trend in China’s AI sector, where commercial readiness and deployment at scale are increasingly valued over standalone technical benchmarks.