hunter-gplpcn
The conventional wisdom about China’s AI industry has been that its models impress and its revenue disappoints. Kuaishou is building a counter-example. In the first quarter of 2026, the company’s Kling AI video-generation business booked revenue of more than 650 million yuan, roughly $90 million, up more than 300% from a year earlier, according to results Kuaishou released this week. By March, the business was running at an annualized revenue rate of nearly $500 million, the company said.
Kling AI began as an experiment in text-to-video generation and grew into one of the most capable video models from China, a competitor to OpenAI’s Sora and to models from Google and ByteDance. Kuaishou, best known for short video and livestream e-commerce, launched the product in 2024 and has pushed it aggressively since, both at home and through an international version aimed at Western users. In February, the company released the Kling 3.0 series of models, updating capabilities, product structure and application scenarios, according to the company.
The numbers show a business at an inflection point. Revenue more than tripling in a year is uncommon even in the young AI-app economy, and the March run rate suggests acceleration: the company’s first-quarter revenue, annualized, sits below the $500 million March figure, which means the business is growing within the quarter as well as across years. Analysts read the pattern as compounding month by month rather than quarter by quarter, a signal that adoption is broadening rather than peaking. Kling 3.0, released in February, appears to be converting users into paying customers.
Kling sells to two audiences. Consumers pay monthly membership fees for generation credits, using the tool to make short videos for social platforms, presentations and personal projects. Businesses buy API access to generate video at scale — advertising creatives, e-commerce product clips, marketing content — and this is where revenue is compounding fastest, according to people familiar with the business. The combination of consumer subscriptions and enterprise API sales is becoming the standard shape of Chinese AI monetization.
Analysts said the results strengthen the case that China’s AI application layer is monetizing faster than many expected. The country’s app companies benefit from a dense mobile ecosystem, a consumer base accustomed to paying small monthly fees, and cloud infrastructure priced well below Western equivalents. Doubao, Kimi and other Chinese AI apps have pushed subscription models this year, and Kling’s numbers give the market a concrete proof point that consumers will pay for AI generation.
Video generation is also the most expensive AI workload to serve, which makes the revenue figure only half the story. Each minute of high-resolution generated video consumes far more compute than text tasks, and margins depend on model efficiency, pricing discipline and scale. Kuaishou runs its own training and inference infrastructure, which gives it cost control that startups renting cloud capacity lack. The economics also work differently for a platform company: Kuaishou can feed generated video directly into its own ecosystem — short-video creators, livestream sellers, advertisers — so every AI video made on Kling can help drive engagement or commerce on Kuaishou itself.
For Kuaishou, AI is still a small share of the top line. The company’s core business — short video, livestreaming, e-commerce — generates the vast majority of revenue, and chief executive Cheng Yixiao has said AI is the company’s growth bet for the second half of the decade. The Kling numbers, disclosed inside the group’s quarterly report, are the first time investors can see that bet in yuan terms rather than promises.
The company has also pushed Kling overseas, where video generation is priced in dollars and competition with Sora, Veo and other international models is direct. International growth adds a currency hedge and a larger addressable market, analysts said, though it also brings copyright, regulatory and content-moderation questions that domestic operations do not raise.
Competition is intensifying on every side. ByteDance’s Jimeng, Alibaba’s video models and international rivals are all chasing the same market, and quality gaps have narrowed quickly. Kling’s advantages are distribution — Kuaishou’s platform reaches hundreds of millions of users — and the practical applications that come with it: e-commerce sellers on Kuaishou already use generated video for product demos, and advertisers are testing AI-generated creatives at scale.
The trajectory will be tested by the costs. Kuaishou has not broken out AI profitability, and analysts said the compute bill will determine how much of the revenue growth survives contact with the income statement. If Kling’s revenue keeps compounding at the current rate, the business will move from a curiosity inside a short-video company to a meaningful contributor to its financials. Analysts said the next quarters will show whether the March run rate was a spike or a trend. Chinese technology shares have broadly rallied this year as investors re-rated the sector on AI, and Kuaishou’s disclosures give that story a revenue foundation. On this quarter’s evidence, China’s AI app economy has found a model that works — and Kuaishou is proving it with numbers.
