Erik Voorhees Predicts AI Will Generate Entire Movies in Minutes by 2030

Erik Voorhees predicts that by 2030, a person could speak a sentence and have AI turn the idea into a film in about a minute. He also expects people to be able to render an idea as a video of any length or as a 3D world in virtual reality. Voorhees says it is easy to underestimate how quickly the technology is advancing. He expects AI tools to become so powerful that building without them will be difficult, a shift he regards as both scary and empowering.

Venice AI gives users access to text and image models from Anthropic, OpenAI, and open-source providers, with video generation increasingly part of the offering. After Venice launched its video studio, it grew to about 30% of platform usage. Users currently make videos of 10, 20, or 30 minutes by generating clips and combining them, a practical example of AI-assisted production that is distinct from Voorhees’s prediction of a film generated from a spoken sentence in about a minute.

The Lumara Film Festival received roughly 700 to 750 submissions, mostly films eight to twelve minutes long. Voorhees said the quality of AI video, the tools available, and artists’ participation had all improved substantially in the six months before the festival. Many entries, in his account, were complete stories with characters, backgrounds, and advanced editing, made by amateurs working alone or in small teams. Some creators had experience in film, motion graphics, or special effects; he said $10,000 of credit on Venice could make a full movie.

Voorhees described the festival’s submissions as evidence that AI video creation is not limited to isolated visual experiments: entries could bring together story, characters, setting, and editing. The range of participants included both amateurs and people with relevant production experience. A festival judge, he said, had created what he called the first AI video to receive an MPAA rating and go into theaters. The Rect Award recognizes satire or comedy about financial markets or crypto.

Privacy is a central part of Venice’s design. Voorhees says the company does not store user content, which remains in the user’s browser or app and is inaccessible to Venice. He argues privacy should be the default, contrasting that approach with major AI companies that store prompts, which he says could be accessed by company staff, governments, or hackers. The distinction is consequential for users sharing prompts or creative material, although the synopsis does not establish how other services handle every kind of content or access.

Voorhees also argues that restrictions on major platforms can frustrate creators and limit artistic freedom. He says Venice users have thanked the company for offering tools without refusing or scolding them. His case for AI creation therefore combines the anticipated speed and breadth of the tools with questions about who controls access to them, how much privacy users retain, and whether platform rules constrain what people can make.