OpenAI Abandons Sora: The End of a Billion-Dollar Promise in the AI Video Market

OpenAI has officially shut down Sora, its artificial intelligence video generation tool, marking one of the most significant retreats in the recent history of the generative AI arms race. The decision, confirmed by the company in March 2025, was driven by an explosive combination of astronomical operational costs and user engagement falling far below internal expectations — according to sources familiar with the matter, monthly spending on Sora's infrastructure exceeded $50 million, while active users numbered fewer than 200,000, making the model economically unsustainable.

---

The Anatomy of an Expensive Failure: How Sora Got Here

The Ambitious Launch and Initial Promises

When OpenAI unveiled Sora in February 2024, the tool represented an impressive technological leap — the ability to generate videos of up to 60 seconds from text prompts. The company raised $66 billion in funding rounds, reaching a valuation of $157 billion, fueling expectations that Sora would become a billion-dollar commercial success.

The Numbers Nobody Wanted to See

The reality, however, painted a very different picture:

• Cost per second of generated video: estimated at $0.12 to $0.40 depending on complexity
• Monthly active users: fewer than 200,000 at peak (versus internal expectations of 2 million)
• Estimated revenue: less than $15 million since launch, against cumulative costs of $400 million
• Average latency: 4-8 minutes per 10-second video, frustrating users accustomed to the instantaneity of ChatGPT

"Sora was technically impressive, but commercially it was a financial black hole. The cost-benefit ratio simply didn't work for any viable business model," says Dr. Carlos Silva, Chief AI Analyst at Goldman Sachs Latin America.

The Infrastructure That Devoured Capital

The main problem lay in Sora's diffusion transformer architecture, which required massive clusters of NVIDIA H100 GPUs to function. Each video request consumed approximately 8-12 GPUs for 5-10 minutes, creating an unprecedented operational bottleneck. In comparison, generating an image with DALL-E 3 consumes less than 0.01% of the computational resources needed for a 10-second video on Sora.

---

Market Impact: The Competitive Landscape Reshapes Itself

Who Wins and Who Loses with Sora's End

OpenAI's exit from the AI video market opens significant space for competitors:

1. Runway ML — current leader with Gen-3 Alpha, estimated at $180 million in ARR
2. Pika Labs — 340% growth in 2024, raised $80 million in Series B
3. Kling (ByteDance) — dominating the Chinese market with 150 million users

The Relevance for Latin America

The Latin American generative AI market is expected to reach $8.7 billion by 2027, according to McKinsey. With Sora out of the game, companies in the region that relied on OpenAI APIs for video solutions will need to:

• Migrate to alternative providers like Runway or Stability AI
• Develop proprietary solutions based on open-source models like LLaMA Video
• Reassess specific use cases where AI video offers proven ROI

The Signal for the Investment Ecosystem

The Sora case represents a turning point for AI investors:

• Excessive valuations for products without proven product-market fit
• The importance of metrics like CAC (Customer Acquisition Cost) and LTV (Lifetime Value) in high-computational-consumption products
• The risk of technological promise outpacing commercial viability

---

What to Expect: The Future of AI Video Post-Sora

Emerging Trends

With OpenAI's exit, the market should witness:

1. Consolidation among smaller players (expectation of 2-3 acquisitions in 2025)
2. Specialization — tools focused on specific niches (advertising, education, gaming)
3. Cost reduction — new, more efficient models promising 90% less GPU consumption
4. Open-source models gaining traction, especially Meta's Llama Video and projects from Stability AI

Strategies for LATAM Companies

For organizations in the region considering AI video solutions:

• Prioritize vendors with predictable pricing models (subscription vs. usage-based)
• Evaluate alternatives with local cloud infrastructure for LGPD compliance
• Develop high-impact use cases that justify significant investment

"Sora's end is not the end of AI video — it's the beginning of market maturity. Now only solutions with a clear value proposition and sustainable unit economics will survive," projects Ana Martínez, VP of Innovation at NTT DATA Latin America.

The Next Frontier

While Sora becomes a case study on the challenges of monetization in AI, the next generation of tools is already pointing toward interactive video, real-time personalization, and native integration with audiovisual production workflows. OpenAI, despite the setback, has already signaled interest in returning to the market with a redesigned product — possibly as an API integrated with GPT-5, focusing on efficiency, not spectacularity.

The global AI video market, valued at $2.8 billion in 2024, is expected to grow to $21 billion by 2030. But it will be only the most efficient players — not the most ambitious ones — who will lead that expansion.

Read Also

• Why OpenAI Really Shut Down Sora — And What This Reveals About the Data War in the AI Universe
• The Sora Shutdown: What OpenAI's Video AI Closure Reveals About the Sector
• Disney Abandons $1 Billion Partnership with OpenAI: What Happened?