The most important investor in China's AI boom isn't a venture fund chasing the next model. It is a state-backed vehicle with a balance sheet measured in tens of billions, and it just picked a side in the video-generation fight. The winner it backed isn't a lab — it's the app inside a video platform most Western users have never opened.
A fund built to be patient
The National AI Industry Investment Fund (国家人工智能产业投资基金) was formally established in January 2025 with registered capital of RMB 60.06 billion (≈ US$8.5 billion / HK$66 billion) — often written as RMB 600.6亿元. It was formed by the National Integrated Circuit Industry Investment Fund Phase III (大基金三期) together with Guozhitou (Shanghai) Private Equity Fund Management, and its controller traces to the Shanghai State-owned Assets Supervision Commission, with the Ministry of Finance as the ultimate majority owner through the Big Fund.
The Ministry of Industry and Information Technology publicly confirmed the fund's launch and its roughly ¥60 billion (≈ US$8.5 billion) scale as part of China's push to turn artificial intelligence from a research topic into industrial infrastructure. Unlike typical venture vehicles, its mandate is explicitly "invest early, invest small, invest frontier" — patient capital aimed at the full chain from compute hardware to models to applications.
For context on scale: the national fund is small next to the Big Fund's semiconductor war chest, but its remit is broader — it can write into software and applications, not just fabrication. That flexibility is exactly why a video model qualified.
The Kling (可灵) injection
On August 31, 2026, Kuaishou Technology announced that the national fund had injected RMB 1.4 billion (≈ US$197 million / HK$1.54 billion) in cash into Beijing Kling (北京可灵), the independent operator of its video-generation large model (大模型) "Kling AI (可灵AI)." The stake was about 1.14%. A separate strategic investor, Charoen Pokphand Group's robotics arm, put in roughly US$19.3 million (≈ RMB 131 million).
The round priced Beijing Kling at a US$15 billion pre-money valuation, about US$18 billion post-money, with Kuaishou retaining roughly 68.33% and continuing to consolidate the unit. The national fund and CP's arm both took contractual repurchase (put) rights — a structural protection that signals this is strategic co-investment, not a passive bet.
Why a video model, and why now
Kling (可灵) is one of the few Chinese generative-video products with genuine global reach, competing head-on with OpenAI's Sora and Google's Veo in the imagination of creators. For a state fund whose thesis is national capability, backing a consumer-facing large model (大模型) with export potential is a notable departure from the usual diet of chips and "hard tech."
It also fits a pattern. The same national fund is part of a wider mobilization of state capital into AI — alongside the Big Fund's semiconductor mandate and a separate national venture-capital guidance fund designed to mobilize roughly ¥1 trillion (≈ US$141 billion) through regional sub-funds. The Kling stake is the most visible sign yet that Beijing's "patient money" is now willing to sit inside a loss-making, fast-moving consumer AI business.
What the deal does not say
The 1.14% slice is symbolic as much as financial — a US$197 million ticket in an US$18 billion company is a flag-planting move. It tells the market the state is comfortable being an anchor in generative AI (生成式AI), even at premium valuations. What it does not tell us is the fund's expected holding period, its governance role, or how aggressively it will deploy the remaining tens of billions of dry powder across the model layer.
Honest limitations
This article draws on Shanghai Securities News (中国证券网) reporting of Kuaishou's announcement and MIIT statements; the fund's size and structure are corroborated by the official enterprise-registry filings cited in that coverage. I have not independently audited Beijing Kling's valuation or unit economics, and the pre/post-money figures come from the company's disclosure as reported. The piece covers only this single investment and the fund's mandate; it does not assess Kling's product quality versus Sora or Veo, regulatory risk around AI-generated content, or how the fund's other deployments may differ. No investment recommendation is made.
What readers can do now
- Watch the national fund's next tickets: if it keeps buying into consumer large models (大模型) rather than only chips, the center of gravity in Chinese AI capital is shifting upstream toward applications.
- For competitive intel, treat Kling (可灵)'s funding as a proxy for where Chinese generative video is headed — track its model releases against Sora and Veo rather than its marketing.
- Global founders should note that China's state capital now co-owns a flagship consumer AI at a ~US$18 billion mark; cross-border partnership or competition with that vehicle is a 2026–2027 reality worth mapping.