Sanxingdui: Future Chronicles has pushed generative AI into a 100-minute theatrical feature. Its first three days at the box office point to a different challenge: turning lower-cost production and technical novelty into audience demand.
China has reached a notable point in the development of AI filmmaking. On October 23, Sanxingdui: Future Chronicles opened in Chinese cinemas, becoming the country’s first feature produced using AI technology to receive a theatrical release permit from the National Film Administration.
The science-fiction animated film runs for 100 minutes. Its production team spent about two years developing the project, generating more than 1.2 million base images and putting the full film through more than 100 iterations.
All of its characters are original digital creations rather than digital replicas of real performers. AI was used extensively for image generation and execution, but the project was not made without a conventional creative team. Nearly 100 people were involved, with roles covering writing, directing, production, art, cinematography, lighting, sound and editing. Creative decisions, according to the filmmakers, remained with the human team.
That distinction matters. Sanxingdui is less a demonstration that a machine can independently make a movie than an attempt to reorganise parts of feature-film production around generative tools.
The film was produced through the Boka Movie Cloud Studio, with Chinese generative-video company ShengShu Technology’s Vidu providing technical support. For a technology more commonly associated with short videos and individual generated shots, maintaining characters, environments and visual continuity across a 100-minute film presented a substantially different production problem.
The result has now reached commercial cinemas. The audience response, however, has been considerably more modest than the scale of the technical experiment.
Maoyan Professional data showed that Sanxingdui had grossed about RMB 1.16 million ($163,000) after its first three days. On October 25, its occupancy rate was about 1.1%, with an average of roughly 1.1 admissions per screening.
Those numbers are early, and one film cannot establish whether audiences will accept or reject AI-produced features as a category. They do, however, expose a gap that is likely to matter as studios put more generative technology into commercial production: making images more efficiently and creating demand for those images are separate problems.
A Future Crisis With Its Answers in the Past
The film itself places AI at the centre of that discussion.
Sanxingdui: Future Chronicles takes the archaeological culture of Sanxingdui in Sichuan and builds a science-fiction adventure around it. Recognisable objects including the bronze sacred tree, large standing figure and bronze masks appear in hyperreal digital form, while the filmmakers extend the archaeological setting into imagined ancient palaces, ritual spaces and a future world.
The story begins with humanity facing a crisis after becoming increasingly dependent on algorithms. A mysterious and incomplete warning from space eventually points back towards Sanxingdui and a series of ancient symbols and artefacts.
The adventure follows Jiang Cheng and his companions as they search for six Sanxingdui relics needed to complete the warning, while uncovering a connection involving astronaut Xiling and the ancient Shu world. The broader mystery revolves around seven ancient symbols associated with Sanxingdui, linking a technological crisis in the future with a civilisation more than 3,000 years in the past.
That gives AI an unusual double role in the project. It is part of the production system used to create the film and part of the subject the film is trying to explore.
But the theatrical release also tests something that cannot be answered by the production process alone. Audiences are still being asked to buy a ticket for a 100-minute story.
The method used to generate the images may attract attention, particularly for a first-of-its-kind release. Over a feature-length running time, however, the same familiar requirements remain: characters, narrative, pacing and a visual experience capable of sustaining audience interest.
The Cost Argument
For the film industry, one of the clearest attractions of generative production is cost.
Producer Quji Xiaojiang has said that Sanxingdui cost more than RMB 10 million to make, while estimating that producing comparable imagery through conventional live-action methods could have required RMB 800 million to RMB 900 million.
That comparison is the producer’s estimate rather than an independently established production budget. Even so, the potential reduction in the cost of creating large environments, crowds and effects-heavy sequences explains why studios are continuing to explore the technology.
Bona is already moving forward with another AI-driven science-fiction project, Unplanned: Guanghan Mission. The film has entered production and is currently working through early material, shot preparation and technical testing.
The direction of investment therefore has not been determined by the opening box office of Sanxingdui alone.
There is a straightforward industrial reason for that. If generative tools can reduce the resources required to produce imagery that would otherwise depend on large sets, extensive visual-effects work or expensive location production, they could change the economics of certain kinds of filmmaking even without replacing the traditional production process as a whole.
Sanxingdui provides an early example of that possibility at feature length.
What it has not yet demonstrated is a corresponding change on the demand side.
Lower production costs can make a film cheaper to supply. They do not make a cinema ticket more valuable to an audience.
That distinction is particularly visible in Sanxingdui’s first weekend. The project crossed several technical and institutional thresholds: a feature-length running time, a theatrical permit, nationwide cinema distribution and a production workflow built extensively around generative AI. Yet its first three days produced only limited theatrical demand.
It is too early to turn that result into a verdict on AI filmmaking. Sanxingdui is one film, with one story, one marketing campaign and one release strategy. Its performance cannot show whether audiences will embrace or reject future films made with similar technology.
The narrower conclusion is more useful.
Generative AI is moving beyond demonstrations and short-form experiments into the practical production of feature films. Sanxingdui: Future Chronicles has taken that transition further than most Chinese projects before it.
But lowering the barrier to making images does not lower the barrier to winning an audience.
For studios experimenting with AI, the first problem is increasingly technical and economic. The second remains much older: making a film people want to see.
Enjoyed This Story? Subscribe to EntertainLens for more entertainment insights.
