Zhang Yimou Discusses AI's Impact on the Film Industry: How Bona, China Film, and Shanghai Film Are Betting on AI Movies

On November 7, at a dialogue event at the Golden Rooster International Film Festival, Zhang Yimou spoke about the pressure AI is bringing to the film industry.

He believes that concerts, stage plays and other live performances depend on being there in person, whereas films present images, so they will be hit more directly by AI. He has observed that at airports and on high-speed trains, many people are scrolling on their phones, watching short videos and short dramas. Such content keeps taking up viewers' time; some AI short videos are not only free to watch, but platforms even give viewers red envelopes for watching them.

This technology has also entered the production process of professional filmmakers. Zhang Yimou revealed that his own "Three-Body" project has already selected more than a dozen post-90s people to form an AI team to assist with visual design.

Bona Film Group has already moved on to its next AI film. According to a November 8 report by Securities Daily, Bona confirmed that the AI-native science fiction film "Unplanned: Guanghan Operation" began shooting in October and is scheduled for release in 2027. Previously, "Sanxingdui: Future Past" entered theaters on October 23.

According to a November 4 report by Caijing, Jiang Defu, producer of "Sanxingdui: Future Past," said the film's actual investment was in the range of ten million yuan. As of November 2, Maoyan data showed its box office at 1.678 million yuan.

For Bona, the experiment in production and delivery has been completed; how much audiences are willing to pay to watch remains another question. Perhaps only with the next film can this experience become a production method the team continues to use. It may be precisely that the company hopes to accumulate experience through these projects so that the next AI film will be easier to organize and produce.

Shanghai Film Group is trying to use AI to develop comic IP. On September 6, the AI web film "Mangosteen Knife," co-produced by Shangying Yuan, launched on Tencent Video. At the subsequent earnings briefing, Shanghai Film stated that the team has already built a production process around this work, from script development to delivery of the finished film, and hopes to apply the relevant experience to the development of more IP.

China Film Group has also included AI works in its investment scope. The second phase of the "China Film Young Filmmakers Program," announced in April this year, has a total scale of 1 billion yuan, of which China Film plans to invest 500 million yuan from its own funds. The investment targets include original films by young filmmakers as well as AI films or short films, and the funds can be used for development, production, and promotion and distribution.

This China Film plan reveals another consideration of film companies: after new tools become widespread, works worth investing in may emerge outside the original circle of partners. Companies need to find people who know how to use these tools early and establish cooperation with them, rather than waiting until a work becomes popular and then fighting over teams the market has already noticed.

Production companies hope to reduce the risk between investment and recovery, copyright-holding companies hope to increase development opportunities, and investment companies also want to find new projects with more controllable costs.

In preparing for AI, film companies cannot focus only on how much money is saved on each shot. Lower costs can expand the room for projects to survive, but whether the investment can ultimately be recovered still depends on whether the works produced have an audience. What they are striving for now is to learn as early as possible how to do business with this set of tools.

Who will the double-edged sword of technological equality strike?

Traditional production companies will face more price competition from small teams.

In the past, some complex images required division of labor among many people and long production times, and manpower and schedule were important bases for quotations. After AI takes on part of the work, small teams also have the opportunity to complete tasks they could not take on before. When clients have more choices, the original quotations will be affected.

Ethan, head of the Yu Guang AI brand, said in an interview with other media that advertising content that previously required 30 to 50 people to collaborate can now be completed by an AI team of 4 to 5 people. He also mentioned that there are already many studios and individual creators in the market taking on low-priced projects, and price competition among peers is very fierce.

After such teams enter the market, clients have new objects for price comparison. If the same production requirements can be completed with fewer people and lower quotations, traditional companies that continue to quote based on the original manpower and schedule may lose orders.

Senior directors will also have to compete with more newcomers holding works for projects.

Traditional film projects usually raise money first and then form a crew to start shooting. Before investors see the finished film, they need to refer to what the director has made in the past and what projects the team has completed. Seniority at this stage can help creators gain trust, while newcomers need to spend more effort persuading others.

Lin Yuncheng previously worked in new media operations for ten years and had not touched AI generation tools before April this year. He completed the 19-minute AI science fiction short film "Under the Sand Curtain" in more than forty days, with computing costs of about four to five thousand yuan, excluding personal labor and equipment investment.

The short film launched on Douyin on August 15. According to a report by relevant media on October 13, the work has received more than 2.78 million likes and nearly 1.5 million reposts. After the work became popular, cooperation invitations from advertisers, advertising agencies and MCNs came one after another. A creator who previously worked in new media operations gained an audience through his own work and also began to be noticed by potential partners.

With such works, newcomers can directly show how they design images, shape characters and tell stories. Partners also have an additional basis for judgment and can decide based on the work whether to continue contact. People who originally lacked a track record have the opportunity to enter the scope of project consideration, and the competitors senior creators need to face will also increase.

There are already examples like "Niu Lai" of newcomers entering theaters on the strength of their works. Director Xin Yumeng previously studied art landscape and spent five years completing this animated film, and explicitly stated that no AI was used in the entire film. If AI can shorten some production stages, more creators may be able to try, and the industry will also see more new faces beyond conventional credentials.

Grasping the tools early can help creators produce works faster and gain attention. As the tools become widespread, more and more people will be able to produce similar images. Once an effect that once required repeated trial and error becomes something others can also do skillfully, it becomes hard for the leader to continue winning collaborations on that basis alone.

This means that after gaining attention with a single short film, creators still need to deliver follow-up works. If they only repeat images they have already shown, the opportunities they can compete for will become increasingly limited.

Technological democratization has opened the door to creation, and it has also brought a growing number of competitors. Whether established teams, veteran directors, or newcomers who have just attracted attention, all need to use their next works to compete for opportunities.

What else do cinemas still have

What Zhang Yimou worries about is that after creative opportunities increase, audiences' time may not necessarily flow to cinemas.

The asset cinemas hold is a mature channel for monetizing films. Distribution companies send films to cinemas in various locations; cinemas arrange screenings and sell tickets, and revenue is then distributed to all parties in the industry chain according to agreements. How much upstream is willing to invest and which directors to hire depends to a large extent on judgments about box office. A director's past box office performance will also affect the chances of obtaining investment and distribution support next time.

Works have new channels for making money, and audiences also have more choices that do not require buying a ticket. This competition does not require a short video to surpass a film; as long as users finish one video and still want to watch the next, an evening that could originally have been used to watch a film may be consumed.

In addition, with one phone per person, audiences are also more accustomed to watching according to their own time and preferences. Even if they want to communicate with others, they can post a comment or forward a video clip, without having to agree on a time to go to the same cinema.

The shared viewing and social opportunities cinemas once provided need more concrete appeal to make people willing to go out.

What cinemas need to do next is make audiences feel that a particular screening is worth a special trip.

On November 2, tickets for this year's Golden Rooster Film Exhibition went on sale, and presales approached one million yuan within an hour. Classic films such as Blade Runner and Inception quickly sold out. Re-releases of old films give audiences the chance to see classics again on the big screen, and also give people who love the same film a reason to gather together.

In addition to screening content, cinemas are also trying to provide new viewing experiences through adjustments to equipment and space. On October 26, People's Daily Overseas Edition reported on the "one chair, one cinema" model of the Xiying Wujie XR Cinema, in which audiences wear headsets and sit in rotatable seats to enter virtual scenes. In July this year, the China Film Administration made clear that virtual reality film screening venues with the corresponding permits are treated as regular cinemas.

In such cinemas, audiences need to be present, but the images are presented through devices worn by each person, and the auditorium is no longer organized around a single screen. The experience cinemas sell has extended from sitting in front of a screen watching a story to entering a virtual scene.

The cinema model that has lasted more than a century may be rewritten amid changes at both ends.

For cinemas, what will need to feel irreplaceable in the future is this particular experience of making a special trip to watch.

This article comes from the WeChat public account "Guduo Network Film and Television" (ID: guduowlj), author: GuDuo Guduo Editorial Department, published by 36Kr with authorization.