OpenAI has shifted its popular Sora video generation app from a free model to a paid system for active users, marking a significant pivot as the company grapples with what it describes as “completely unsustainable” economics for the AI-powered tool.

The change, announced Thursday by Sora CEO Bill Peebles, allows users to purchase bundles of 10 additional video generations for $4 through Apple’s App Store after exceeding their daily free limits. The pricing applies across all subscription tiers, including free users as well as Plus, Teams, and Pro subscribers.

GPU Resource Constraints Drive Monetization

Currently, free, Plus, and Teams plan users can generate up to 30 videos daily, while Pro subscribers receive 100 free generations. However, Peebles warned these limits will likely decrease as the platform grows.

“We were quite surprised by how actively our power users want to use Sora, and the economics are currently completely unsustainable,” Peebles wrote on X. “We thought 30 free generations per day would be more than enough, but clearly we were wrong!”

The company cited GPU resource limitations as the primary reason for the changes. “Eventually we’ll need to reduce the number of free generations to enable growth (otherwise we won’t have enough GPUs for it!), but we’ll be transparent when that happens,” Peebles added.

The admission reveals OpenAI’s challenge in balancing user demand with computational reality. Video generation requires dramatically more processing power than text or image generation, with each video consuming GPU resources equivalent to potentially thousands of text responses. At scale, these costs become prohibitive without corresponding revenue.

Vision for “Sora’s New Economy”

Beyond addressing immediate sustainability concerns, OpenAI outlined plans to build a broader “Sora new economy” that would enable creators and rights holders to monetize their content. The company intends to launch a pilot system where users can pay additional fees to use copyrighted characters or celebrity likenesses in videos, with revenue potentially shared with original creators and rights holders.

“We envision a world where rights holders can decide for themselves whether to charge extra for favorite characters and people appearing,” Peebles explained. “Soon we’ll launch a monetization pilot, prioritizing those people and companies who were early to join the platform.”

This creator monetization concept addresses two challenges simultaneously. First, it provides a legal framework for using copyrighted material in AI-generated content—a persistent concern across the AI industry. Second, it creates revenue streams for rights holders who might otherwise oppose AI tools that enable unauthorized use of their intellectual property.

The success of this model depends on rights holder participation. If major studios, agencies, and creators embrace the system, it could establish a template for sustainable AI content generation that respects intellectual property. If they reject it, OpenAI may face ongoing legal battles over unauthorized use.

Trademark Dispute Complicates Rollout

The monetization announcement arrived amid legal challenges related to Sora’s “Cameo” feature. Chicago-based company Cameo, which offers personalized celebrity video messages, filed a trademark infringement lawsuit against OpenAI in US federal court Tuesday, arguing the feature name creates consumer confusion and threatens the company’s business model. Cameo CEO Steven Galanis called the situation an “existential threat” to his company’s operations.

The trademark conflict highlights risks in naming AI features after existing services. Whether intentional or coincidental, the choice of “Cameo” for a celebrity-likeness feature directly invades the namespace of a company whose entire business centers on that concept. The lawsuit outcome could force OpenAI to rebrand the feature regardless of trademark validity.

From Cameo’s perspective, the threat seems genuine. If users can generate synthetic celebrity videos through Sora instead of paying for authentic messages through Cameo’s platform, the business model collapses. Whether courts view this as trademark confusion or legitimate competition will significantly impact both companies.

Rapid Adoption Exceeds Expectations

Since launching as a standalone app on September 30, Sora has reached over 2 million downloads in the US and Canada alone, quickly climbing to first place in Apple’s App Store. The app’s rapid popularity exceeded OpenAI’s expectations, forcing the company to balance user demand with computational costs amid trademark disputes in the competitive AI video generation market.

The 2 million downloads suggest strong consumer interest in AI video generation, though download numbers don’t necessarily correlate with sustained usage. Many users likely experiment with the technology out of curiosity without becoming regular creators. The subset of power users generating dozens of videos daily represents OpenAI’s immediate monetization target.

The first-place App Store ranking demonstrates Sora’s cultural moment despite limited availability and constrained free generation limits. Competing tools from Runway, Pika, and others haven’t achieved comparable mainstream penetration, giving OpenAI a window to establish market leadership before competitors catch up.

Pricing Strategy and Market Positioning

The $4 for 10 videos pricing translates to $0.40 per video—potentially expensive for casual experimentation but reasonable for serious creators. The pricing also ensures OpenAI captures revenue from its heaviest users while maintaining free access for occasional use cases.

Competitors will likely watch OpenAI’s pricing experiment closely. If users accept $0.40 per video as reasonable, other platforms may adopt similar pricing. If backlash occurs, competitors can undercut OpenAI or maintain more generous free tiers as differentiation.

The tiered system creating different daily limits across subscription levels adds complexity but enables granular monetization. Free users get a taste of capabilities, Plus/Teams users receive moderate access, Pro subscribers get substantial daily generation, and everyone can purchase additional capacity when needed.

Sustainability Challenges in AI Video

OpenAI’s frank admission about unsustainable economics reveals broader challenges facing AI video generation. Unlike text models where inference costs have dropped dramatically, video generation remains computationally expensive with limited room for efficiency gains using current architectures.

The GPU constraint Peebles mentioned isn’t just about money—it’s about physical hardware availability. Even with unlimited budget, acquiring additional GPUs faces supply chain limitations from Nvidia and data center capacity constraints. OpenAI must balance serving existing users with leaving headroom for growth.

The “eventually we’ll need to reduce free generations” warning suggests current limits represent temporary generosity until user base grows further. Users accessing Sora today enjoy more favorable terms than future users likely will, creating incentive to establish usage patterns before restrictions tighten.

Whether OpenAI’s paid option successfully addresses sustainability concerns depends on adoption rates among power users, efficiency improvements in video generation models, and whether the creator economy materializes as envisioned. The next several months will reveal whether Sora’s economic model can support the demand it has generated.