Kuaishou restructures Kling AI as independent unit, eyes $20B valuation and external funding
The AMW Read
Kuaishou's spin-off updates the generative media player map with a major new corporate structure and validates the capital-compression pattern in AI video; highly significant for the generative media segment but does not resolve a foundational open debate.
Kuaishou restructures Kling AI as independent unit, eyes $20B valuation and external funding
Kuaishou has announced plans to restructure Kling AI (可灵AI), its AI video generation business, into an independent business unit, evaluating external financing and a potential Hong Kong IPO. The company is reportedly targeting a $20 billion valuation for Kling AI and seeking approximately $2 billion in funding, with Tencent named as a potential investor. Kling AI reached roughly $500 million in annualized recurring revenue (ARR) as of early 2026, up from $240 million at end of 2025, making it one of the fastest-growing AI video platforms globally.
Why it matters: Kuaishou's move mirrors a broader capital-structuring pattern in China's AI industry — spinning off capital-intensive AI assets into separately financed entities to shield the parent company's margins while allowing the AI business to raise dedicated funding. Baidu did the same with Kunlun Core (昆仑芯), its AI chip unit now pursuing a Hong Kong IPO, and Alibaba has been preparing T-Head (平头哥) for a similar carve-out. The Kling AI restructuring updates the Segment 01 player map — where Kuaishou joins ByteDance (Seedance), Alibaba (HappyHorse), and Runway in the AI video trench — and signals that generative media's compute cost curve is pushing even public-company-backed labs to seek external capital rather than internal cross-subsidy. This also exemplifies the "fastest-ARR-ramp" pattern: Kling AI went from $60 million to $500 million in ARR inside 12 months, yet its compute costs are rising so fast that internal funding was straining Kuaishou's profitability, triggering a 14% stock drop on the parent's FY2025 earnings. The split is a vote that generative video's capital needs are closer to frontier-model scale than to consumer app scale.
The move effectively validates a structural thesis: AI video generation requires compute budgets large enough to reshape corporate finance. With OpenAI recently sunsetting Sora due to inference costs and ByteDance reportedly spending $23 billion on AI infrastructure in 2026, Kuaishou's carve-out is a pragmatic acknowledgment that Kling AI's compute intensity no longer fits inside a single B2C company's P&L. The explicit mention of Tencent as a potential investor — alongside Tencent's own admission that its video model HunyuanVideo 1.5 lags the front rank — suggests the acqui-licensing or strategic-investment pattern may be at play, where a hyperscaler buys into rather than builds the category leader.
