AI Long Drama 'Hou Xi You Ji' Lifts Market Cap by 13.2 Billion Yuan

On August 31, ‘Hou Xi You Ji,’ described as China’s first AIGC long drama series, was simultaneously streamed and broadcast in prime time on Mango TV and Hunan Satellite TV. Planned as 30 episodes of about 40 minutes each, the first five episodes were released first. No live-action actors or filmed footage were used, and all visuals were generated with ByteDance’s Volcano Engine large models Seedance 2.0 and Seedance 2.5. According to reporting by Liu Shuhan of Huxiu, the project is positioned as a touchstone for the higher quality and longer form AI video that streamers and production companies have pursued for two years. While AI short dramas have circulated in large numbers, long-form has been seen as a difficult area to conquer. This broadcast stimulated the capital market as the first case to break through that barrier.

China’s First AI Long Drama Stimulates the Market

In the five trading days after the broadcast, the share price of Mango Super Media, under Hunan Broadcasting System, rose about 50% cumulatively. The increase in market capitalization reached about 13.284 billion yuan. The rally was not limited to the company, spreading to film and media stocks as a whole. In the same period, Bona Film, Hualu Century, Zhongguang Tianze and others were bought up to their daily limit. On September 7, on the first trading day after the final episode aired, Mango Super Media rose another 4%. It closed at 22.1 yuan per share, with market capitalization reaching 41.33 billion yuan. Chief director Li Dongsheng spoke about the market reaction as follows.

It was unexpected, but on reflection it was only natural

He cited the industry’s need for a new topic as the reason.

The industry desperately needed a new topic

He also expressed the view that the high cost of conventional video production has put pressure on both streamers and production companies. The enthusiasm did not last long. After peaking on September 7, the company’s stock corrected for three consecutive days. It closed at 17.77 yuan on September 11, down nearly 20% from the high. Market capitalization shed about 8.1 billion yuan from the peak, and trading volume shrank to less than half of the peak. With the first five episodes finished and the next release scheduled for the winter holiday, the market’s initial test completed a cycle from overheating to cooling in less than two weeks.

Breakdown of the Low Cost of 900,000 Yuan

per Episode

The total budget for ‘Hou Xi You Ji’ is said to be about 900,000 yuan per episode. That works out to 20,000 to 30,000 yuan per minute. Compared with conventional long dramas and animation, it is an order of magnitude lower. The production period was also short. The concept was finalized in March this year, the project was formally launched in April, and it aired in August. This contrasts with conventional live-action dramas, which generally take one to two years. The difference in cost structure is striking in action scenes. For about three minutes of action scenes, conventional methods would have required about 4 million yuan. With AI production, by contrast, it reportedly cost only 30,000 to 40,000 yuan. That is one-hundredth the cost. Computing resource costs accounted for about one-fourth to one-fifth of the total. 4K generation cost more than 100 yuan per run. The production team was nearly 100 people in size. About 70 core production staff participated in AIGC-related work. The art team had about 15 people, drawing concept art for scenes and main characters and setting the aesthetic direction for the work as a whole. The AI team was divided into seven groups, each consisting of three to four people. Seven AI directors were assigned, one for each group. Many of the core talent came from film universities, communication universities, and art universities. Conventional roles such as scriptwriting, editing, and production remained. The original work is a supernatural novel of the same name from the late Ming to early Qing dynasty. Set long after Sun Wukong’s departure, it depicts Sun Xiaosheng, a small stone monkey raised on Flower-Fruit Mountain, growing as he defeats enemies and setting out again on the journey to the West. Viewer reviews were polarized. Some praised the texture of the Oriental aesthetic created by AI. Others pointed out stiff acting, slow pacing, and remaining AI-specific unnaturalness.

Technical Challenges Remain in Expressions

and Consistency

The technical constraints are clear. Fine facial expressions of AI-generated characters lack sophistication. Long takes are prone to scene drift. The production team avoided this by using a method that puts only one to three cuts into a single asset. They adjusted camera movements and chose compositions less prone to unnaturalness. One example is the scene of the grown-up Sun Xiaosheng. With the same model and prompts, the small stone monkey period featured lively and natural acting. Grown up in armor, his expression became stiff and his movements awkward. The team suspected the armor was restricting movement, changed the prompts, adjusted expressions, and swapped reference materials. They gave detailed instructions down to how muscles move, but improvement was not easy. It was a problem that took nearly 10 days. The director does not see AI as an extension of conventional tools. He stresses the need to build an entirely new workflow. The AI team’s job was not simply repeating generation. Everyone understood storyboards and cinematography, and took charge of emotional and narrative judgments. The system integrated prompt design, storyboard composition, and material selection. Chief director Li Dongsheng is an experienced documentary director. With more than a decade of experience, he has directed and produced historical documentaries. Around 2020, in a work dealing with ancient history, he tried a method combining painting and CG. Looking back later, he says the idea was close to image-to-video generation. The turning point was a roughly 150-minute documentary produced in 2025. To fully recreate the history of Taiwan Island, it required numerous naval battles and historical reenactment scenes. Conventional live-action filming or CG would have been expensive and time-consuming. During preparation, Mango TV’s technical team showed him the AIGC platform under development, which became the catalyst. Initially skeptical of AI tools made by conventional media, his judgment changed after seeing the model’s generation results. With the release of Seedance 2.0 early this year, he decided to take on the long-form challenge. At almost the same time, Mango TV established the AIGC Innovative Content Center, and ‘Hou Xi You Ji’ was the first strategic-level project after its establishment.

Challenge to Establish a Reproducible

Industrialized Process

‘Hou Xi You Ji’ is positioned as a strategic touchstone for Mango’s AIGC business. Its mission is three-layered: verifying technical feasibility, establishing a standardized workflow, and exploring IP-derivative business. Long-term value is seen not in the popularity of a single episode but in the accumulation of a reproducible process. On the business side, there have been early responses. It reportedly gained traction in increased male viewers, advertising, and derivative products. But the commercial cycle is not yet mature. Industrialized production routes and a virtuous commercial cycle are not yet fully established. Whether the approach, honed over months by a team of nearly 100, can be applied as-is to other works remains unverified. Li Dongsheng calls himself a radical. He expressed the view that AIGC will be involved in more than 90% of video works in the next three years. Whether this prediction holds depends on standardization of processes and stability of cost-effectiveness. With the next release scheduled for the winter holiday, sustained quality and stable delivery will be tested. After the capital market overheating cooled, what the industry must verify is not topicality but reproducibility.

Editorial Opinion

In the short term, we expect video companies to bring forward follow-up AIGC long-form projects. The sharp rise and fall of Mango Super Media shares showed that topic-led buying easily attracts short-term funds. Over the next three to six months, we assess that the ability to secure computing resources and manage processes will separate winners from losers. If crude copycat projects increase, we expect viewers to become more selective. In the long term, we expect structural changes in production costs to alter the profitability of streaming businesses. If the level of 900,000 yuan per episode is reproduced, it will become easier to test mid-sized IPs, and we assess that opportunities will expand for new directors and local stations. Meanwhile, quality barriers remain, such as expressions and long-duration consistency, and we expect the division of labor between live-action and generation to be redefined in one to three years. We believe a reshuffling of jobs and skills is unavoidable. The question from the editorial team is what to measure reproducibility by. We believe the criterion will be not only cost and schedule, but whether a different team can deliver the same quality on a different work. We also consider it a point of debate whether viewer acceptance should be judged by short-term view counts. Without disclosure of processes and third-party verification, will this achievement end as a one-off success?

References

• “ 90万一集的AI长剧，如何炸出芒果130亿市值？ “, by 刘纾含 — 虎嗅网, 2026-09-11T23:32:36.000Z (ARR)
• Source URL: https://www.huxiu.com/article/4890512.html?f=rss