Luma joins forces with Wonder Project to form Innovative Dreams: AI video company starts making movies

While most AI video companies are still competing over "whose model generates smoother clips," Luma made a radically different choice: On April 16, it announced the establishment of Innovative Dreams in conjunction with the Wonder Project - a Production Services Company that combines generative AI with traditional film production. From "selling models" to "making movies", this is a profound industry depth.

Structure and intent of the joint venture

Wonder Project is helmed by well-known film and television producers and focuses on combining cutting-edge technology with narrative content. The establishment of Innovative Dreams means that Luma's technical assets - Ray series video models, Uni series brand models, Luma Agent - will directly serve professional content production processes such as movies and advertisements, and explore a hybrid production model of "generative AI pre-visualization + traditional live shooting/post-production".

From "providing tools" to "intervening in the industry chain"

The strategic intention of this move is not difficult to interpret: AI video companies have begun to deeply intervene in the content industry chain, rather than just "providing generation tools." Model capabilities are upstream, and content production and distribution are downstream. Luma chose to go downstream, using a joint venture production service company to verify the "AI native production" business model, and at the same time find higher-quality real production scene feedback for model capabilities.

From an industry perspective, this path is not unique to Luma: Runway cooperates with media groups, and Kling cooperates with the film and television industry. The trend is highly consistent - AI video competition in 2026 has extended from "model capabilities" to "content production and distribution" level. Luma's choice is particularly radical, because it is not just a "cooperation", but a direct establishment of a production company. For domestic AI video manufacturers, this reminds us of a reality: when models tend to be homogeneous, whoever can take the lead in opening up the industry closed loop of "from prompt words to finished films to accounting" will have the right to define the next stage.

Several directions worth tracking in the future:

1. Innovative Dreams’ first batch of projects: Can it produce AI-native content with box office/commercial value?
2. Implementation of pre-visualization workflow: The adoption rate of AI pre-visualization in real film pre-production.
3. Cooperation with studios/streaming media: Whether the joint venture will bind Hollywood or domestic film and television resources.
4. Model feedback closed loop: Movie-level production data feeds back the training of subsequent versions of the Ray series.