Kling AI annualized revenue reaches US$240 million

Background

AI video generation turns a text prompt, an image or an existing clip into new footage. Running these models is compute-heavy, and until recently the field was judged mainly by demo reels and benchmark scores. Kling AI is Kuaishou's video generation product, and it is one of the services trying to turn that capability into a subscription and enterprise business rather than a demonstration.

What Kuaishou announced

On 13 January 2026, Kuaishou said that Kling AI's monthly revenue passed US$20 million in December 2025.^[1] On an annualized basis that is US$240 million.^[1] The company framed the number as an annualized revenue run rate, usually shortened to ARR: it takes a single month's revenue and multiplies it by twelve to show what a full year would look like if that pace held. ARR is a snapshot of current momentum, not a forecast, and it moves with any month-to-month change in sales.

Kuaishou also placed the figure in time: the milestone came 19 months after launch.^[1] That interval matters because it sets the revenue against the product's age. Building a paying customer base for a generative model is a different problem from building the model itself, and the announcement was as much about commercial adoption as about the technology.

The usage behind the number

Kuaishou reported that Kling had more than 60 million creators and more than 600 million generated videos.^[1] It also said the service worked with more than 30,000 enterprise users.^[1] Those three figures describe different parts of the business: individual creators generating clips, the total volume of output those creators produced, and organisations that use the product through commercial arrangements rather than casual use.

A product push in the same period

Kling AI held an "Omni Launch Week" in early December 2025.^[1] During that launch, Kling Video 2.6 added simultaneous audio-visual generation, meaning sound and picture are produced together rather than assembled in a separate step, and Kling Image O1 and Kling Digital Human 2.0 were released in the same launch.^[1]

Kling Video O1 was described as supporting multimodal video editing and video referencing, with a subject library intended to improve character and object consistency.^[1] Consistency is a familiar obstacle in generated video: if the same character or product has to appear across several shots, small variations between clips become visible, and the footage has to be regenerated or fixed afterwards. A subject library is one way to hold that identity steady across generations.

Where the revenue is meant to come from

Kuaishou listed the target sectors for Kling AI as marketing, e-commerce, film and television, short plays, animation and gaming.^[1] These are fields that already produce large amounts of visual material and routinely need many variations of the same idea. Marketing and e-commerce need product shots and promotional clips at volume; film, television, short plays and animation need storyboards, concept frames and pre-visualisation before expensive shooting begins; games need concept art and promotional footage. In each case the value comes from producing candidates quickly and then choosing among them, which is why repeatable characters and objects matter as much as raw image quality.

Why the figure is notable

For a young category, revenue is a different kind of evidence from a demo. A monthly revenue figure and an enterprise user count say that customers are paying repeatedly, not just trying a tool once. The announcement gives a concrete reference point for how large the paid market for AI video had grown by the end of 2025, and it ties that growth to a specific product line with named versions. It does not, on its own, say how profitable the business is: serving video generation is computationally expensive, and the announcement concerned revenue rather than costs or margins.