Kuaishou's AI Video Spinoff Unveils New Model in ByteDance Chase

NextFin News - Kling AI, the artificial-intelligence video unit Kuaishou Technology is spinning off, has unveiled a new model as its race against ByteDance intensifies - and the launch lays bare the central tension of the deal. The Beijing-based company is pushing its AI video arm toward an independent listing at a target valuation of roughly $20 billion, a price tag that values a business contributing less than 1% of group revenue at about two-thirds of the parent company's entire market capitalization.

Kuaishou confirmed in a May filing that its board is evaluating a restructuring of Kling AI's assets and business, which may involve external financing. By July, the plan had moved from speculation to a filed fact: a group of investors including Alibaba, Tencent and Baidu agreed to inject more than 19 billion yuan ($2.8 billion) into the unit, valuing it at $15 billion on a pre-money basis and diluting Kuaishou's stake to about 68% from full ownership. The round is capped at 20.45 billion yuan, or roughly $3 billion. The new model is the product proof point behind that valuation, and it arrives as the market's initial enthusiasm for the spinoff has faded.

The Launch and the Valuation Gap

The latest Kling model builds on the 3.0 series the company unveiled earlier in 2026, which supports native 4K output, clips of up to 15 seconds at 60 frames per second, and multi-shot storyboarding that sequences up to six camera shots inside a single generation. The company says the models integrate text, image, audio and video into one workflow - text-to-video, image-to-video, reference-to-video and in-video editing under a single multimodal architecture. The newest release extends that trajectory as Kuaishou races to keep pace with ByteDance's Seedance, which launched on ByteDance's CapCut editing platform in March and gained roughly 45 million monthly active users during the Chinese New Year period, nearly four times Kling's 12 million.

The commercial traction behind the launch is real, and unusually so for a category where most Western competitors have struggled to monetize. Kling AI reached $100 million in annualized revenue just 10 months after its June 2024 launch, hit $240 million by December 2025, and crossed roughly $500 million in annualized revenue by April 2026. In the March quarter alone, the unit generated 650 million yuan in revenue, more than quadrupling from a year earlier, and in the second quarter revenue exceeded 850 million yuan, up more than 200% year over year. About 70% to 75% of that revenue comes from overseas markets, particularly North America. As of December 2025, the company said it served more than 60 million creators, had generated more than 600 million videos, and had partnerships with more than 30,000 enterprise clients.

The market's first read on the spinoff was enthusiastic. On May 12, shares of Kuaishou surged as much as 11% at the open, briefly touching HK$57.40 intraday after reports emerged that the company was in discussions with investors including Tencent to raise about $2 billion at the $20 billion valuation. By the close, the spinoff's proposed valuation was equivalent to roughly 68% of the listed parent's market capitalization. But the enthusiasm has not held. Kuaishou closed at HK$30.02 on September 25, down 1.57% on the day, and the stock is trading near its 52-week low of HK$29.14 - roughly half its May high. The AI story is no longer enough to offset weakness in the core short-video business.

The Valuation Paradox: Pricing the Option, Not the Earnings

The $20 billion target is not a valuation derived from Kling AI's current earnings. It is a bet on what Kuaishou's core business is no longer worth. Kling generated 340 million yuan in revenue in the fourth quarter of 2025 against group revenue of 39.6 billion yuan - a 0.86% share. A $20 billion price tag implies a price-to-sales multiple that the parent company, with its plateauing user growth and decelerating advertising revenue, cannot command.

This is the structural logic of the spinoff, and it cuts both ways. On one side, a standalone Kling AI can raise capital at an AI-era multiple, fund its own compute buildout, and attract talent with equity that is not tied to a short-video operator's earnings. On the other side, the transaction effectively strips Kuaishou's most valuable growth option and leaves public shareholders with a mature advertising and live-streaming business whose multiple has been under pressure all year.

The math explains why the stock initially jumped and then gave back the gains. Investors priced the optionality on the announcement; they repriced the sum-of-the-parts on reflection. A subsidiary generating less than 1% of revenue cannot, on its own earnings, support a $20 billion mark. The valuation only works if Kling AI's revenue keeps compounding at its current pace and if the parent's core business does not deteriorate further.

The ByteDance Race: Product Parity, Compute Asymmetry

The new model arrives in a market where the technical gap between the leading Chinese video generators has narrowed faster than the capital gap. ByteDance's Seedance 2.0 matches Kling on native 4K and multi-shot storytelling while pushing harder on multimodal reference inputs - up to nine reference images plus video and audio conditioning. Seedance 2.5 extends native generation to 30 seconds, beyond Kling's 15-second ceiling. Both companies are racing on features, but they are not racing on equal footing.

ByteDance has raised its 2026 AI infrastructure spending plan to more than 200 billion yuan (about $30 billion), an increase of at least 25% from its earlier plan, with a larger share of the budget directed to domestic AI chips. Kuaishou's total capital expenditure for 2026 is expected to reach approximately 26 billion yuan, an increase of about 11 billion yuan from 2025, with the added spending primarily allocated to computing power for the Kling model and other foundational models. That is a disparity in financial firepower of roughly seven to one.

This is where the spinoff becomes a strategic necessity rather than a financial engineering exercise. Kling AI cannot out-spend ByteDance from within Kuaishou's balance sheet. As a standalone entity with its own funding round and a listed currency, it can finance the compute arms race without dragging down the parent's earnings per share. The alternative - funding AI capex inside a slowing advertising business - is what has punished Kuaishou's shares all year. Kuaishou ended 2025 with 104.9 billion yuan in available funds and generated 26.7 billion yuan in operating cash flow, so it has capacity - but every yuan spent on Kling inference and training is a yuan the market now scrutinizes as a drag on near-term margins.

Second-Order Effect: The Core Business Is Being Repriced as a Cash Cow

The first-order read of the spinoff is that it unlocks value in Kling AI. The second-order effect is that it redefines what Kuaishou itself is. After the transaction, Kuaishou will own about 68% of a high-multiple AI company and 100% of a low-multiple short-video platform. The market will increasingly price the parent as a holding company whose job is to harvest cash from live-streaming and advertising and funnel it into an AI bet it no longer fully controls.

That is a less exciting equity story than the one Kuaishou has told for the past two years. The company built its valuation on the promise that AI would transform its core recommendation engine, advertising targeting, and e-commerce conversion. A spun-off Kling AI still benefits Kuaishou through its retained stake, but the strategic integration - the reason investors paid a premium for the AI optionality inside the listed vehicle - becomes harder to underwrite.

The revenue mix makes the tension concrete. Kuaishou's second-quarter revenue rose 1.4% year over year to 35.5 billion yuan, with adjusted net profit of 3.9 billion yuan and an 11% margin. In the company's own words, the AI push is already delivering:

"In the second quarter of 2026, we continued to unlock the strategic value of AI and delivered solid financial and operating performance amid a complex macroeconomic environment," said Cheng Yixiao, Kuaishou's co-founder, chairman and chief executive officer. "We accelerated the integration of AI across our content ecosystem, online marketing, e-commerce and internal organizational enablement, enhancing both user experience and commercial efficiency."

Cyclical or Structural: This Is a Structural Repricing, Not a Sentiment Dip

The decline in Kuaishou's shares is not a cyclical fluctuation that will revert on its own. Three structural forces are at work, and none is mean-reverting.

First, user growth in China's short-video market has plateaued. With roughly 700 million monthly active users spending more than 130 minutes per day on its services, Kuaishou is at the mature end of the adoption curve. Growth will come from monetization, not new users - and monetization growth has already slowed to the low single digits.

Second, the advertising and live-gifting revenue model faces regulatory and behavioral headwinds that do not self-correct. Live-streaming gifting, once the profit engine, has been constrained by tighter rules on tipping and by changing consumer behavior. That is a regime change, not a cycle.

Third, the AI video category is undergoing a permanent shift in how value is captured. The winners will be the companies that can fund the compute arms race and build distribution, not the ones with the best single model. Kling AI's $500 million annualized revenue and 60 million creators give it a credible position - but only if it can stay funded. The spinoff is the mechanism that makes that possible.

The implication is uncomfortable for holders of the parent stock: even if Kling AI succeeds, Kuaishou's shares may not fully participate. A successful spinoff means the AI value accrues to a separate listed entity, while the parent retains a 68% stake that the market will discount as a holding-company position. The structural call is that the sum-of-the-parts discount will persist until either Kling AI lists and the discount crystallizes, or the parent's core business reaccelerates - and there is little evidence of the latter.

The Counter-Thesis: The Market Is Wrong to Discount the Option

The strongest case against this reading is that the market is mispricing optionality it does not understand. Kling AI is not just another AI video startup. It is the AI video company with the largest non-Anglophone user base, a verified commercial model, and revenue growing more than 300% year over year. By comparison, OpenAI's Sora generated an estimated $2.1 million in total lifetime revenue from in-app purchases before its shutdown, according to mobile-data firm Appfigures. The gap between technical spectacle and commercial durability is where Kling AI's valuation premium lives.

Analysts who see value in the transaction point to the same math from the other direction. Morgan Stanley has said the transaction would unlock significant value for Kuaishou, and Haitong International has projected Kuaishou's share price could reach HK$84 following the spin-off - nearly triple the recent level. J.P. Morgan projects Kling AI could reach $1.3 billion in annualized revenue by the first quarter of 2027, which at a 40x revenue multiple would imply a $52 billion valuation for the unit and a HK$73 target for the parent.

This counter-thesis has force. If Kling AI's revenue doubles again over the next 12 months and the Hong Kong listing prices at a premium, the retained 68% stake alone could be worth more than Kuaishou's current entire market capitalization. The market would be making the classic mistake of valuing a call option at zero because the underlying is not yet profitable.

The answer to the counter-thesis is not that it is wrong, but that it is early. Optionality has value only if it is exercised, and exercising it requires capital that Kuaishou cannot provide from its own balance sheet without damaging earnings. The spinoff is the bridge to that capital - but bridges are valuable only if you reach the other side. Until Kling AI lists and the market can price the shares directly, the option remains illiquid and discounted. The counter-thesis wins only if the listing happens on schedule and the AI video market does not consolidate around ByteDance before then.

What to Watch

The forward path splits into three time horizons, and they point in different directions.

In the short term, sentiment will track the listing timeline. Kuaishou has confirmed it is evaluating a restructuring that may involve external financing, and reports point to a Hong Kong listing as early as 2027. Any delay, or any sign that investors are balking at the $20 billion target, will pressure the shares. The falsifying signal for the structural-discount thesis is specific: if Kuaishou files for Kling AI's Hong Kong listing within the next two quarters and the IPO prices at or above $20 billion, the market's current discount is wrong and the parent's shares should re-rate higher.

In the medium term, the key metric is Kling AI's annualized revenue trajectory. J.P. Morgan's $1.3 billion projection for the first quarter of 2027 is the benchmark. If Kling AI gets there, the valuation is supported by revenue, not speculation. If annualized revenue stalls below $800 million, the $20 billion target becomes untenable and the parent's retained stake will be marked down.

In the long term, the question is whether the AI video market consolidates around one or two platforms. ByteDance's distribution advantage through Douyin and TikTok is formidable, but Kling AI's overseas revenue concentration - 70% to 75% of the total - gives it a distribution channel ByteDance cannot easily replicate in Western markets. The structural winner will be the company that combines model quality with funded compute and non-China distribution. Kling AI has two of the three; ByteDance has all three.

The base case is that the spinoff proceeds, Kling AI lists in 2027 at a valuation between $15 billion and $20 billion, and Kuaishou's shares recover modestly as the holding-company discount narrows. The upside case is a hotly contested listing that prices above $20 billion, triggering a re-rating of the parent toward HK$84 as analysts have projected. The downside case is a delayed or downsized listing that leaves Kuaishou holding an illiquid stake in a cash-burning unit while the core business continues to decelerate.

The beneficiaries are clear: Kling AI's new investors, who bought in at $15 billion pre-money; creators and enterprise clients, who gain a better-funded, faster-innovating platform; and patient holders of Kuaishou who can wait for the listing to crystallize the discount. The exposed are holders who bought the AI narrative inside the listed vehicle and now face a multi-year wait for that narrative to become a tradable asset.

Kuaishou is not spinning off Kling AI because it has won the AI video race. It is spinning it off because it cannot afford to keep funding a race it cannot afford to lose - and the market is only now pricing in what that admission costs.

Explore more exclusive insights at nextfin.ai.