OpenAI terminated its Sora video generation service on March 24, 2026, after the product burned through an estimated $15 million per day in inference costs while generating only $2.1 million in total lifetime revenue from in-app purchases. The shutdown highlights the fundamental economic challenges facing AI video generation at consumer scale, according to analysis by Revolution in AI.
Bill Peebles, OpenAI’s head of Sora, acknowledged on social media that “the economics are completely unsustainable.” The service’s cost structure stemmed from video generation’s computational intensity—each second of video requires rendering hundreds of individual frames with spatial reasoning about motion, lighting, physics, and temporal consistency.
Forbes reporting cited across multiple outlets estimated Sora’s peak daily inference costs at $15 million, translating to approximately $5.4 billion annually just for server operations. Against this backdrop, mobile intelligence firm Appfigures reported that Sora’s in-app purchase model generated $2.1 million in total lifetime revenue—a gap the credit-based purchasing system could never bridge.
User engagement data revealed a sharp decline that preceded the shutdown. While Sora’s standalone app launched in September 2025 and reached the top of iOS App Store’s Photo and Video category within one day, monthly downloads peaked at 3.33 million in November 2025 before plummeting 66% to 1.1 million by February 2026. Monthly active users similarly peaked in December 2025 and declined through early 2026.
KeyBanc Capital Markets analyst Justin Patterson noted in a research note that even with OpenAI’s resources, Sora failed to attract and retain an engaged audience. The service’s peak performance of 3.3 million monthly downloads across all platforms contrasted sharply with ChatGPT’s 900 million weekly active users, underscoring the challenge of building a sustainable video generation business at current compute costs.
