Kuaishou's AI drama engine now ships 470 titles daily—actors optional
The Kling platform has industrialized China's $6.9 billion short-drama market, dropping production costs 90% and eliminating crews, sets, and human talent from most new releases.
The story
Kuaishou's Kling AI has crossed from research demo to industrial content pipeline. The platform now generates 470 short dramas daily[1], each typically running 60–90 episodes at roughly one minute per episode, with production costs down 90 percent from traditional shoots. The workflow is end-to-end synthetic: text-to-script, script-to-storyboard, Kling's native video generation for footage, AI voice cloning for dialogue, and algorithmic scoring for music. Human actors, directors of photography, and location scouts are optional at best. In a market valued at $6.9 billion and growing, the marginal cost of a new title has collapsed toward the price of GPU cycles and electricity. This is not a pilot program. We're tracking daily output that already rivals the combined slate of most traditional studios, and the velocity is only constrained by compute capacity and platform moderation bandwidth. The economic structure of the short-drama segment—single-digit-dollar production budgets recovered through in-app purchases and ad impressions—already rewarded speed and volume over prestige. Kling's toolchain eliminates the last friction: coordinating humans. The result is a Cambrian explosion of formulaic content optimized for algorithmic recommendation rather than creative differentiation. Titles are A/B tested at launch, and underperformers are killed within hours while breakout concepts spawn dozens of clones by week's end. The second-order effect is more disruptive than the headline efficiency gain. Kuaishou now controls the entire value chain from model weights to distribution platform, and the feedback loop is instantaneous: viewer engagement data trains the recommendation engine, which in turn shapes the generative model's reinforcement fine-tuning. Competitors without vertically integrated inference and distribution infrastructure face a structural cost and iteration-speed disadvantage. The stock dropped 3 percent on the catalyst day, likely reflecting concern that the model compresses margins across the ecosystem faster than it expands Kuaishou's own monetization—advertisers pay per impression, and when supply goes parabolic, CPMs crater unless demand scales in lockstep.
The rest of this story is for subscribers.
Including Our Take, the Tailwinds & headwinds framing, Connections across the FOBI roster, and What should you do.
Already subscribed? Sign in →
