Netflix debuts generative AI in El Eternauta to create a building collapse sequence, speeding VFX creation by 10× and reducing costs.
Companies specializing in developing video AI models and tools frequently discuss working with Hollywood studios to facilitate specific workflows.
On Thursday, Netflix announced that it has initiated the utilization of AI in its productions of films and television programs.
During the company’s post-results conference call, Ted Sarandos, the co-CEO, claimed that the platform had “the very first GenAI final footage to appear on screen” in an Argentine program called “El Atonata.”
He observed that Netflix’s internal production group collaborated with producers to produce a scene depicting a building collapsing using AI.
The scene was completed at a rate that was ten times quicker than it would have been with conventional visual effect tools, according to Sarandos, and at a lower cost.
“We continue to be convinced that AI presents an extraordinary opportunity to assist creators in producing films and series that are more cost-effective and superior.”
There are creator tools that AI propels. This is an example of individuals engaging in real labor with improved tools.
The advantages of pre-visualization and shot-planning work, as well as visual effects, are already being observed by our creators in production. “In the past, advanced visual effects such as de-aging were exclusively available to large-scale projects,” he stated during the call.
Greg Peters, Netflix’s co-CEO, stated that the company utilizes generative AI in various applications, such as personalization, search, and advertising. He also noted that the company intends to implement interactive ads in the latter half of the year. The company implemented AI-powered search earlier this year.
In the second quarter, Netflix reported a profit of $3.13 billion and revenue of $11.08 billion, representing a 16% increase from the previous year.
It observed that users viewed over 95 billion hours of content in the first half of 2025, with non-English titles comprising one-third of all user views.
