Is AI short drama really profitable: the gap between surging production capacity and a profit margin of about 5%

When AI drives down the cost of producing digital content and pushes up massive supply, the micro-drama and AI comic-drama tracks begin to fall into a cycle of "infinite content supply and absolutely scarce attention." Beneath the appearance of prosperity, the industry's overall profit matrix is decoupled from capacity expansion.

Although the entire industry is witnessing a supply explosion, the proportion that truly crosses the break-even line may be only about 5%.

Output Soars

From September to October 2026, the penetration rate of AI dramas and comic dramas among leading short-drama apps at home and abroad rapidly rose to 90%, and the switch in content formats is nearly complete.

The market size of domestic micro-dramas (including comic dramas) grew from 37.4 billion yuan in 2023 to 100 billion yuan in 2025, and is expected to reach 121 billion yuan in 2026; AI dramas and comic dramas, meanwhile, are set to leap from about 16.8 billion yuan in 2025 to more than 40 billion yuan in 2026, a year-on-year increase of about 138%.

The expansion trend on the supply side is astonishing. In August alone, 74,313 new AI short dramas were launched on Douyin, averaging about 2,400 per day, more than 110 times the 655 in August 2025; in the first half of the year, a cumulative 221,900 new titles were released.

The industry-wide figure is equally staggering: from January to September 2026, 490,000 micro-dramas were launched nationwide, 13 times the full-year total for 2025.

Micro-dramas are upgrading from a supplementary content format to infrastructure-level content.

However, the prosperity on the production side is ruthlessly diluted by the hit-show funnel. Of the 221,900 new dramas, about 12,000 surpassed 10 million views, accounting for 5.8%; about 2,900 reached the break-even line, accounting for 1.3%; and only 1,055 exceeded 100 million views, accounting for 0.48%. Only about one in every 77 new dramas recoups its cost.

This set of funnel data is the coldest footnote in this wave. Capacity has never been the barrier; the ability to consistently produce hit shows is. A business model that bets on probability through capacity is, in essence, working for channels and traffic.

Profit distribution shows a Matthew effect. The top 10% of dramas can achieve profit margins of 50% to 80%, 42% are marginally profitable, and 21% are loss-making; the estimated industry average gross profit is -3,962 yuan per drama; the only hedge is that one hit drama exceeding 100 million views can cover the cost of about 15 dramas.

By conversion, about 70% of dramas are in the marginally profitable range or below, and the industry average line remains negative. "Profitability" is still just a hypothesis awaiting verification for the vast majority of players; the real deciding factor is the product of hit rate and unit cost, not the number of releases.

Midstream players anchored to capacity have already been cleared out first. The number of active production contractors shrank from 1,216 to 698, a drop of 42%, and contract production unit prices were rapidly compressed.

Costs Fell by 90%

Over the past year, the per-minute production cost of AI comic dramas fell from nearly 15,000 yuan to about 1,500 yuan, a drop of about 90%, and the cost per title was compressed from 500,000 to 800,000 yuan for live-action short dramas to 50,000 to 240,000 yuan.

The collapse of the cost curve is the primary driving force behind this round of explosion. When the production threshold is leveled by a 90% decline, the focus of competition inevitably shifts from capacity to traffic.

The cost structure itself already points to where the money goes: token costs account for about 40%, manual adjustment about 45%, and fees paid to IP holders about 10% to 15%.

In other words, production fees do not settle as assets for producers but continuously flow out along three directions: computing power, labor, and IP; producers are more like handlers of expenses than retainers of profit.

The market is expanding, and the loss-making scope is also expanding. Citi estimates that about 95% of producers may be in a loss-making state, and only about 5% may be profitable. The coexistence of "a 40 billion yuan market" and "a 95% loss-making scope" precisely shows that the cost-reduction dividend has not stayed in the production segment.

Profits along the industry chain are being redistributed. Traffic bidding under oversupply pushed the industry's daily traffic-buying spend in 2026 to 150 million to 170 million yuan, with traffic costs up 100% year on year; contract production unit prices were pressed down to 300 to 500 yuan per minute, leaving contract manufacturing with a net margin of only 5% to 8%.

Traffic buying squeezes production, and platforms squeeze contract manufacturers; bargaining power along the industry chain is accelerating toward the traffic end.

Scarce segments, meanwhile, capture high gross margins. At the model layer, Seedance has a gross margin of about 70% and a daily computing-power share of about 80%; in the first five months of 2026, online-literature IP licensing transactions reached 1.8 billion yuan, up 120% year on year; in the overseas segment, FlareFlow's net margin reached 60%.

AI short dramas compress the added value of midstream production to about 14%, while upstream IP and screenwriting rise to 15%, and downstream platforms and distribution rise to 48%—platforms simultaneously control traffic allocation and settlement data, further strengthening their bargaining power.

Among the comprehensive costs of micro-dramas, platform and traffic-buying fees account for as much as 82.5%, while production, actors, and scripts combined account for only about 14%. What AI can compress is precisely that portion of less than 20%—this is where the answer lies to why "90% cost reduction" and "thin profits across the entire chain" hold true at the same time.

In other words, "production has become cheaper" and "production has become profitable" are two different things.

Cost reduction is universal and cannot constitute a barrier for any producer; what is truly scarce is the three toll booths: models, traffic, and IP.

Domestic Market Competes on Channels, Overseas Market Competes on Profit

The slope on the demand side is still rising. In August, the short-drama app industry's MAU reached 470 million, up about 78% year on year; Hongguo's MAU exceeded 400 million, and its DAU surpassed 200 million in September; Tencent's WeChat short dramas saw MAU exceed 100 million.

Usage time and stickiness are rising in tandem. Hongguo's MAU reached 430 million in September and its DAU reached 200 million, with average daily usage time per person at 166 minutes, up 36% year on year.

Deduplicated users across the entire industry reached 851 million, accounting for 66.7% of active mobile internet users, with about 133 million added in the past four months.

The outline of the user base is now clear: nearly two-thirds of active mobile internet users are consuming short dramas, and the incremental dividend is beginning to narrow. Hongguo has announced that it will take the coin feature offline starting November 1 and bid farewell to subsidy-driven user acquisition—the platforms' competitive focus is shifting from user acquisition to revenue sharing and monetization efficiency.

Channel competition is distributing a phased dividend. Tencent's share is expected to rise from 10%–20% to around 30%, and Tencent and Kuaishou offer higher rebates than Hongguo and ByteDance; during the National Day holiday, ByteDance's single-day short-drama spend exceeded 300 million yuan, and the industry expects it could reach 400 million yuan by the Spring Festival, with institutions viewing Q4 through the first half of next year as a growth window.

For content producers, the marginal improvement in rebates and revenue sharing is the variable most worth tracking over the next two quarters.

At the same time, the overseas narrative is shifting from "grabbing territory" to "calculating profit."

In Q3, overseas short-drama app downloads exceeded 800 million, up about 7% quarter on quarter, while in-app purchase revenue reached $850 million, up 27% quarter on quarter—revenue growth significantly outpacing downloads.

The stratification of per-user value determines the playbook. In the first half of 2026, overseas short-drama app in-app purchase revenue was about $1.27 billion, with Chinese overseas apps contributing about 93%. In North America, each download corresponds to $4.3 in in-app purchase revenue, versus only $0.5 in Asia.

The top players have already shifted into a profit-preservation phase. The top five take 54% of in-app purchase revenue, with DramaBox and ReelShort posting first-half in-app purchase revenue of about $195 million and $188 million, respectively.

DramaWave's September downloads fell 43% year on year, reflecting a deliberate pullback in ad spend, while ByteDance's PineDrama, with its free model, surpassed 100 million downloads in nine months since launch and has about 7 million DAU, staking out the incremental segment.

Commercialization validation is landing in tandem. ReelShort operator CMS achieved profitability in the first half, and Chinese Online's FlareFlow achieved monthly operating break-even.

Kunlun Tech's AI short-drama revenue in September approached $100 million, with ARR exceeding $700 million—going overseas is becoming the highest-margin link in the entire industry chain.

Monetization models are also leveling up, shifting from traffic-based settlement to project-based pricing. The first AI long-form drama to air on satellite TV, "Journey to the West: The Next Chapter," costs about 900,000 yuan per episode, just one-tenth that of traditional animation; two iQIYI AI films generated over 6 million yuan in revenue sharing within ten days of launch, and the platform is offering an additional 20% revenue-sharing subsidy for all-AI content this year.

For producers, the optimal solution within the window is not to expand capacity, but to bind with platforms, lock in revenue sharing, and realize cash flow. In the era of technological equality, value concentrating at both ends—IP and channels—is the most certain trend in this round of distribution.

This article is from the WeChat public account "Gelonghui Finance Hotspots" (ID: glh_finance), author: Godzilla, published by 36Kr with authorization.