AI short-drama supply is surging. Why does Citi prefer platforms, IP and tools?
China released 13 times as many short dramas in the first eight months of 2026 as in all of 2025, but advertising budgets and viewing time have not kept pace with the expansion in supply, putting pressure on producers’ profits. Citi believes returns are more likely to accrue to businesses controlling traffic, quality rights and generation tools; opportunities to cut costs in long-form series and films are constrained by content quality and the costs of existing projects.
Original source：Citi ResearchFirst published on Unhidea：Original source date：
Key takeaway
Citi believes that as AI lowers production barriers and intensifies content competition, traffic gateways, quality IP and generation tools are better positioned to retain returns than content production alone.
How fast are short dramas growing, and why is it harder for producers to make money?
In the first eight months of 2026, China released 430 thousand micro-dramas, 13 times the total for all of 2025; more than nine in ten were AI-produced. Demand is also growing: total time spent on short dramas in August rose 195% year on year, and their share of total online time in China increased from 2.4% a year earlier to 6.6%.
But advertising budgets and user time have not kept pace with expanding supply. From month 1 through July of 2026, advertising accounted for 71% of China’s short-drama market revenue; as content competes for limited advertising and user attention, production costs are falling while producers’ profits are under pressure.
Supply has begun to fall: Citi’s research shows monthly supply declined from a first-half peak of more than 100 thousand titles to about 70 thousand–80 thousand. If producers’ margins do not recover, the bank expects some companies to exit or turn overseas, and domestic supply may continue to decline.
How can platforms gain more bargaining power?
According to industry experts interviewed by Citi, Hongguo Short Drama has a market share of about 60%–70%, with no strong second-place rival yet challenging its position. In 2026, Hongguo scrapped minimum guarantees for AI dramas and switched to revenue sharing, making producers more dependent on their works’ actual performance; new rules effective in September have added compliance costs.
Using 50 million views as the break-even benchmark, Dataeye counted just 2886 new dramas exceeding the threshold in the first half of 2026, or 1.3% of new releases. This title-based measure differs from the company-based estimate in Citi’s expert interviews that “about 5% of producers may be profitable,” but both show competitive pressure.
Citi sees Kuaishou as having the greatest potential to expand its short-drama share, including by strengthening partnerships with Tencent, China Literature and others. A strong platform capable of challenging Hongguo is an important condition for improving producers’ bargaining power.
Why might quality IP be worth more than production capability?
As AI lowers production barriers, Citi expects viewing to concentrate on quality content and IP. In 2025, around 70% of short-drama content supply came from online-novel IP; China Literature ranked first among rights holders by views of AI animated dramas on Hongguo in the first half of 2026, providing a foundation for it to capture adaptation demand.
China Literature’s revenue from short dramas and AI animated dramas exceeded 430 million yuan in the first half of 2026, with a margin of about 10%, driven mainly by hits. Short dramas also compete for time spent reading online novels, so Citi’s view that IP value is rising is accompanied by pressure from audiences being diverted from the company’s online literature business.
How can AI tools turn content expansion into revenue?
Tool revenue is directly linked to production spending: according to experts interviewed by Citi, model-inference token fees account for about 40% of AI short-drama production costs, while IP fees account for 10%–15%. The more active content production is, the more opportunities generation tools have to charge fees.
The opportunity remains constrained by market share. Dataeye expects Seedance to account for 69% of AI-drama foundation-model revenue in 2026, and Kling for 17%; Citi is positive on Kling’s future updates and MiniMax H3’s capabilities, and lists Kuaishou, MiniMax and China Literature as preferred picks.
How quickly can cost cuts in long-form series and films improve platform profitability?
Based on calculations for selected projects, Citi estimates AI content costs just 10%–50% of comparable live-action production, and therefore expects its use to expand into long-form series and films; higher quality requirements, however, will slow that expansion. Bona’s Sanxingdui: Past of the Future is scheduled for release on 23 October 2026 and is seen by the bank as a near-term industry catalyst.
Savings on individual projects take time to feed through to group profits. Both iQIYI and Mango reported second-quarter 2026 profits below Citi’s expectations, while existing live-action projects still account for the bulk of content costs; industry competition also intensified during the period, and total time spent on long-form video fell 16% year on year in August. Citi therefore sees new AI content releases as near-term catalysts, while remaining cautious about a lasting reversal of declining user time.
