In the first half of 2026, AI made producing a micro-drama as effortless as posting a social media update — but it also accelerated the pace of financial losses.

On one side, Douyin launched 221,900 native AI-generated animated micro-dramas, amassing 515.738 billion total views with the annual market size projected to exceed 40 billion yuan and users surging from 120 million in 2025 to 600 million. On the other side sits the sobering reality: only 1 in every 77 new productions recoups its investment, leaving 98.7% failing to reach the break-even point within six months. The Micro-Drama Development Management Measures, effective September 1, will serve as the needle to burst this bubble, forcing the industry to shift from "quantitative expansion" to "qualitative improvement."

AI Lowers Not Costs, But the Threshold for Losing Money

Scale explosion and hit scarcity are occurring simultaneously. The mid-tier production cost for AI animated dramas is merely 800–1,200 yuan per minute, bringing total costs to just one-fifth of live-action productions. This ultra-low barrier unlocked extreme production capacity, yet capacity itself creates no value — only breakout hits do. When supply multiplies a hundredfold in a short period, the traffic allocated to each individual work gets exponentially diluted. Lin Qiwen, Vice President of DataEye, stated in an interview: "Many industry practitioners believe AI lowered the production barrier, but it hasn't made creators more profitable — it's actually made them lose money faster."

New drama releases have been declining slightly month-over-month since March, as industry regulation and platform guidance push the sector from a "batch release, indiscriminate volume" model toward "improved quality with reduced quantity." Future new releases are expected to stabilize at 30,000–50,000 titles. On September 1, the Micro-Drama Development Management Measures (SARFT Order No. 16) officially takes effect. AI micro-dramas will be classified into three regulatory tiers based on investment amount, with 300,000 yuan and 800,000 yuan serving as the two threshold lines: Tier One covers investments of 800,000 yuan or more or special-subject content, requiring certification from the national regulator; Tier Two covers investments between 300,000 and 800,000 yuan for general content, reviewed by provincial broadcasting authorities; Tier Three covers investments below 300,000 yuan, subject to platform review. Article 17 of the Measures clarifies that Tier Three micro-dramas require qualifying broadcast platforms to fulfill content management duties, conducting pre-broadcast review and assigning program numbers. The true power of this regulation isn't blocking the 1.3% of winners — it's holding platforms accountable for reviewing the 98.7% of "money-losing capacity."

Audiences Don't Reject AI — They Reject Repetition

In the summer of 2026, short-video users noticed something: the protagonists in AI-generated micro-dramas all looked increasingly alike. DataEye's report shows that AI humanoid-simulation dramas on Douyin expanded from 12% to 70% of new releases, reaching 70.62% in June. On the viewership side, of the 221,900 new dramas, only 1,055 surpassed 100 million views — a hit rate of just 0.48%. The viewership share of AI humanoid-simulation dramas expanded from 17% in January to 80% in April–May, before contracting to 68% in June due to regulatory and platform guidance. "A thousand faces, one look" is no longer just audience mockery — it's an industry fact confirmed by data.

Type distribution provides the most direct data evidence of this sameness. In the first two months of the year, Douyin's native AI dramas were dominated by 2D and 3D animation. In March, AI humanoid-simulation dramas exploded, accounting for nearly 65% of new releases; by Q2, this share grew from 57% to 70%, steadily capturing 60–70% of supply. 3D animation held steady at around 20%, while 2D animation was squeezed down to 4%. But the viewership story differs: AI humanoid-simulation dramas' viewing share rose from 17% in January to 80% in April–May, nearly monopolizing the screen. By June, however, this figure had contracted to 68%, dropping roughly 12 percentage points from its April–May peak. A scissors gap emerged between quantity and views: AI humanoid-simulation dramas' share of new releases continued climbing (70.62% in Q2), yet their viewership share turned downward. Meanwhile, 3D animation, with 16 times the view growth, captured the share ceded by AI humanoid-simulation dramas, boosting its viewing share from a March low of 6.71% to nearly 30% (29.48%) by June. As the chart illustrates, production volume and viewership are not directly proportional — 3D animation leveraged a stable, modest number of new releases to drive dramatic viewership growth, once again proving that fewer, higher-quality productions can break through the homogenization fog.

In Douyin's first-half top-viewership rankings, the 3D animated drama "Banished to the Border: A Wronged Wife Raises a War God" led by a landslide with 2.127 billion views — a landmark event signaling the end of AI humanoid-simulation dramas' "thousand faces" dominance. The same face — "almond eyes, petal lips, soft-focus filtering" — hit its viewership ceiling in June. Article 34 of the Micro-Drama Development Management Measures states that micro-dramas generated or produced using artificial intelligence technology must have production institutions and broadcast platforms comply with relevant national regulations, adding prominent AI identification labels in every episode. Though this clause appears technical, its essence is to visually mark and distinguish "thousand faces," accelerating audience aesthetic differentiation. Once "thousand faces" are labeled by new regulations and eliminated by audience taste, high-quality technological presentation (such as 3D animation) becomes the breakthrough point.

From Post-Release Complaints to Pre-Release Platform Screening

When "thousand faces, one look" triggered user backlash and 98.7% of capacity became money-losing, platforms had to move first. Douyin and Hongguo's governance actions form a self-purification combination. On the production side, rules are being set: Hongguo recently required AI drama character creation to avoid similarity and eliminate sameness, launching a special campaign targeting "high-frequency AI faces, homogenized content, and material violations." For dramas determined to be "high-frequency reuse," traffic-control measures such as "zero traffic" are implemented. As early as April, Hongguo announced it had intercepted and removed 3,522 low-quality animated dramas within one week; by end of April, the platform declared it had processed over 10,000 low-quality AI dramas involving issues like rough visuals, chaotic plots, vulgar curiosity-baiting, and more.

On the distribution side, risks are being blocked: Douyin's Short Drama Creator Center launched a pre-screening tool on August 3, available to AI drama rights holders and directing institutions, capable of screening for portrait rights and copyright risks before final release. Article 33 of the Measures requires pre-broadcast review in accordance with regulations and synchronized reporting to authorities; for key accounts distributing micro-dramas, a credit evaluation system should be established with focused management. Article 37 further requires: "Broadcast platforms shall periodically review, evaluate, and verify algorithm mechanisms, models, data, and application results, prioritizing recommendation of quality micro-dramas, and shall not use algorithmic models that induce user addiction or excessive consumption." Article 38 mandates a tiered management system and exit mechanism.

Behind the AI drama surge lies the sharp contraction of live-action productions. Hongguo's monthly ranking of new live-action dramas fell from a February peak of 432 titles to 110 by June — a 65% drop — with June popularity rankings down 53%. The reason is clear: starting in April, Hongguo released large volumes of AI animated dramas, crowding out live-action micro-dramas from ranking positions. This reflects an overall user migration — in the first half of the year, AI drama users surged from 120 million in 2025 to 600 million, accounting for roughly 70% of the 851 million total micro-drama users; the remaining 251 million live-action viewers are becoming the squeezed party.

But the policy wind is turning. On May 14, SARFT announced the "Micro-Drama Quality Creation and Dissemination Plan Implementation Plan," with six key platforms committing at least 6 billion yuan to support excellent live-action micro-drama creation and dissemination. Earlier, at April's 13th China Online Audiovisual Conference, Douyin Group had already announced 500 million yuan in special funding to support live-action content innovation and realistic-subject exploration, launching a "Short Drama Creator Center" to promote transparent revenue sharing. With real money entering the field, platforms have pre-positioned supply lines for live-action content in the "post-AI era." Regulatory authorities are also making their moves: entry barriers are rising and compliance costs are increasing — this "one rise, one fall" dynamic frames the direction of industry migration.

The market is large enough, users deep enough, and engagement sticky enough for the industrial foundation to withstand one round of supply-side clearing. From the 121 billion yuan micro-drama market size, to 851 million users, to Hongguo's 357 million monthly active users with 125 minutes of daily usage, each layer of the foundation is thicker than the last. And the fact that AI animated dramas have a hit rate below 0.1% demonstrates precisely that this thick foundation can absorb a reshuffle: the new regulations aren't an emergency brake — they're a lane change on a long, steep slope with deep snow.