In what might be one of the most extravagant examples of AI’s growing costs, OpenAI is reportedly burning through as much as $15 million per day to power its video generation tool, Sora. The tool, which allows users to generate realistic video clips from short text prompts, has exploded in popularity across social media. Yet many of those clips—ranging from dancing penguins to surreal anime skits—are being labeled by critics as “silly,” raising questions about whether the massive spending is truly worth it.
At first glance, Sora represents a stunning leap in artificial intelligence. Users can type a sentence such as “a robot dog surfing on the moon,” and within seconds, the model produces a high-definition, lifelike video. The technology has fascinated creators, marketers, and tech enthusiasts worldwide. However, the scale of Sora’s usage has come at a steep price—one that’s forcing OpenAI to confront the realities of AI economics.
A Costly Creative Playground
According to internal estimates and independent analyses, each 10-second Sora video may cost around one dollar or more in computational expenses. That figure may sound small until it’s multiplied by millions of users generating countless clips each day. With the app’s viral rise, OpenAI could be footing a daily bill in the range of $10 to $15 million just to keep Sora running.
Most of those videos aren’t for professional use. They’re playful experiments, memes, or short artistic tests—creative, yes, but not necessarily commercially valuable. That’s why some industry observers have described the spending as “burning millions a day on silly videos.”
For OpenAI, however, this wave of experimentation might not be entirely wasteful. The company has always operated with the understanding that early adoption often comes at a loss. As with ChatGPT, which once operated at a loss before its subscription tiers stabilized revenue, the goal with Sora appears to be long-term dominance. OpenAI is betting that as costs decrease and monetization improves, Sora will evolve into a sustainable and profitable platform.
The Economics Behind the Burn
Generating high-quality video using AI requires immense computational power. Each clip involves not just image synthesis, but frame-by-frame motion generation, physics simulation, and rendering—all happening across clusters of powerful GPUs. These machines consume vast amounts of electricity and require expensive maintenance.
Even with OpenAI’s access to cutting-edge infrastructure—thanks to its close partnership with Microsoft—the operational costs remain enormous. Running millions of video generations a day demands thousands of top-tier GPUs running nonstop.
Initially, Sora offered generous free access to attract users and build data. But as the costs mounted, OpenAI introduced paid options and limits. Users now receive a smaller number of free daily video generations, after which they can purchase additional credits. Still, many users stay within the free tier, leaving the company to shoulder the bulk of expenses.
The result is a business model that, for now, seems upside-down: massive popularity, but even more massive costs. Some analysts have compared it to the early days of streaming services, where companies spent heavily to attract viewers before finding viable subscription models.
From Innovation to “AI Slop”
The internet has responded to Sora in a mix of fascination and ridicule. While some users create stunning cinematic clips, others flood social media with bizarre or humorous videos—AI cats riding motorcycles, fish walking on land, or surreal parodies of film scenes.
These playful creations have earned Sora a reputation for generating “AI slop”—a term used to describe the flood of low-effort or absurd AI-generated content that clogs digital feeds. Critics argue that OpenAI’s technology is being wasted on internet jokes rather than meaningful artistic or educational projects.
Yet there’s another way to view it. Supporters say Sora’s popularity, even for whimsical content, proves how accessible and engaging the technology has become. Every viral meme, no matter how silly, demonstrates Sora’s creative potential and helps train a generation of users to interact with generative AI.
OpenAI’s Long Game
For OpenAI, Sora represents far more than a viral toy—it’s a critical step toward the company’s broader vision of multimodal AI. While text and image generation have already reached mainstream adoption through ChatGPT and DALL·E, video is seen as the next frontier. Mastering it could position OpenAI as a leader in a market that may soon rival Hollywood and advertising combined.
To reach that goal, however, OpenAI must balance innovation with sustainability. Spending billions annually to power a free or low-cost creative app is not feasible forever. The company is already testing monetization models, including subscription tiers, usage-based pricing, and premium tools for filmmakers and advertisers. If it can successfully convert just a small fraction of Sora’s massive user base into paying customers, the investment could pay off.
The other major priority is efficiency. As hardware improves and OpenAI refines its video-generation model, the cost per video should drop dramatically. This mirrors the trajectory of text and image generation, which were also prohibitively expensive at first but became more affordable as algorithms improved.
The Bigger Picture
The debate over Sora’s spending highlights a deeper issue in the AI industry: the cost of creativity. Generative AI has democratized access to tools that once required professional equipment and years of training. But that democratization comes at a steep financial and environmental cost, as the energy demands of massive AI models continue to rise.
Some experts argue that the current AI boom resembles the early days of the internet—a period defined by reckless spending, rapid experimentation, and uncertain profit models. In that sense, Sora’s $15 million-a-day burn rate could be viewed not as waste, but as the price of innovation.
Still, others see it as a warning sign. If the bulk of that money fuels frivolous content rather than valuable applications, OpenAI risks repeating the mistakes of other tech giants that prioritized engagement over purpose. As the novelty wears off, sustaining such high costs will demand a clearer path to profitability.
The Future of “Silly”
In the end, what critics call “silly” might turn out to be strategically brilliant. Viral, low-stakes content keeps Sora in the cultural spotlight, ensures continuous feedback, and drives data collection essential for model improvement. Every clip—even the absurd ones—helps refine OpenAI’s understanding of how people use AI creatively.
Still, the company faces pressure to prove that Sora is more than a novelty. Whether through partnerships with filmmakers, educators, or businesses, OpenAI will need to show that the platform can deliver real-world value beyond internet humor.
Until then, Sora stands as both a marvel and a money pit—an emblem of the power and the peril of the AI revolution. OpenAI’s multimillion-dollar daily gamble on “silly” videos may look extravagant today, but in the long view, it could either define the next era of creative technology—or become a cautionary tale of digital excess.
