Select Language:
OpenAI’s iOS video generation app, Sora, launched on September 30th through an invitation-only system, has seen rapid growth in just over a month. Within the first week, the app surpassed one million downloads, sparking widespread media attention and a flood of bizarre content. Users created everything from fake security footage and humorous clips featuring deceased celebrities to unsettling home shopping advertisements. Data from AppFigures indicates that by Halloween, October 31st, Sora’s downloads had skyrocketed to around 4 million, with millions of 10-second AI-generated videos produced daily.
Amid this explosive popularity, questions about the financial sustainability of such a massive content push have arisen. Forbes estimates that OpenAI’s daily expenditure on Sora could be around 15 million dollars, translating to an annual cost exceeding 5 billion dollars, or roughly 356 billion RMB. Bill Peebles, Sora’s project lead, acknowledged the current economic model as “completely unsustainable.” Despite this, OpenAI has not disclosed detailed usage metrics or cost breakdowns.
Offering free, unlimited access to such a resource appears to be a high-stakes gamble. While common in the tech world, this approach remains bold. OpenAI is absorbing significant losses to capture market share and establish user loyalty, with the hope that as usage expands, the cost per video will decline sharply. When monetization begins, revenue streams are expected to increase rapidly, balancing out earlier losses.
Analysts from Mizuho Securities, including Lloyd Walmsley, suggest this strategy mirrors traditional internet growth tactics—prioritizing user base and engagement over immediate profit. Walmsley emphasized that advancements in GPU technology could dramatically reduce the cost of video inference, with projections indicating costs could fall to one-fifth of current levels by next year and to one-third by 2027.
This aggressive push appears to be laying the groundwork for future monetization, possibly combining advertising revenue with high-value paid services for professional users like filmmakers and advertising creatives. Additionally, the free access to Sora generates valuable training data—user prompts and generated videos—that can be used to refine AI models, especially since video data remains scarce and difficult to annotate. Such data-driven improvements could enhance OpenAI’s competitiveness and profitability long-term.
Nevertheless, the rapid rise in operational costs has prompted OpenAI to rethink its approach. The company recently announced plans to sharply reduce the free generation quota, acknowledging that the current volume of casual, humorous use—mostly shared among friends—cannot sustain the heavy computational costs. As Sam Altman stated in a recent interview with Stratechery, “Right now, most users are just creating funny memes to send to friends, and no advertising model can support the sheer scale of this infrastructure.” Moving forward, OpenAI aims to balance widespread accessibility with strategic monetization efforts to ensure the platform’s viability.
ChatGPT
Perplexity
Gemini AI
Grok AI
