In a notable shift in the competitive landscape of generative AI, OpenAI has announced the discontinuation of its Sora video generation app, while Google DeepMind continues to push forward with enhancements to its Veo model family. The move highlights the high costs and strategic challenges of scaling consumer-facing AI video tools.
OpenAI’s Abrupt Decision to Shut Down Sora
On March 24, 2026, OpenAI surprised users and industry observers by declaring it was “saying goodbye” to Sora, its standalone AI video-generation app and related professional platform. Launched around September 2025, the app allowed users to create short, hyper-realistic videos from text prompts and featured a social-style feed for sharing content. It quickly gained popularity, briefly topping app store charts, but engagement reportedly declined over time.
The company cited a need to refocus resources amid growing compute demands. Video generation is notoriously expensive due to intensive GPU requirements, and Sora’s operational costs proved unsustainable relative to revenue. OpenAI indicated that the research team would pivot toward “world simulation” efforts, particularly to advance robotics and real-world physical tasks.
The shutdown also led to the dissolution of a high-profile partnership with Disney. Just months earlier, the two companies had signed a multi-year deal—reportedly involving potential equity investment worth around $1 billion—to incorporate licensed Disney characters from Marvel, Pixar, Star Wars, and more into Sora. Disney was reportedly notified with little advance warning, and the agreement has now been terminated.
Users have been advised to export any generated videos before the service is fully wound down. While the consumer app and professional platform are being discontinued, OpenAI has not ruled out continued internal research on video-related technologies.
Google’s Steady Progress with Veo
In contrast, Google DeepMind has been methodically iterating on its Veo video generation models. The latest version, Veo 3.1 (with variants including a lighter “Fast” option), builds on previous releases by introducing richer native audio generation, improved narrative control, better prompt adherence, and enhanced realism in physics and motion.
Key features of Veo 3.1 include:
- Native audio synchronization, encompassing dialogue, sound effects, and ambient noise that matches the visuals.
- Support for reference images to maintain character and style consistency across scenes.
- Flexible aspect ratios, including vertical (9:16) for social media and portrait formats, alongside landscape options.
- Higher-resolution outputs, with upscaling to 1080p or 4K.
- Tools for extending clips, creating seamless transitions, and more precise editing.
These capabilities are integrated across Google’s ecosystem, including the Gemini app (for paid subscribers), Google AI Studio, Vertex AI for developers, and creative tools like Flow for filmmakers. A more cost-efficient Veo 3.1 Lite variant has also been introduced to broaden accessibility while reducing inference expenses.
Google’s approach emphasizes practical applications in advertising, content creation, and enterprise workflows rather than a standalone viral consumer app.
Broader Implications for AI Video Generation
The contrasting paths of OpenAI and Google underscore ongoing industry challenges: explosive compute costs, safety concerns around deepfakes and misinformation, copyright issues, and the difficulty of monetizing flashy consumer tools. While Sora’s brief run captured public imagination with its cinematic quality, its closure serves as a reminder that technical prowess alone does not guarantee long-term viability in a resource-constrained environment.
Other competitors, such as Kling, Luma Dream Machine, and Runway, continue to evolve, but Google’s deep integration with search, advertising, and cloud services gives it structural advantages in scaling responsibly.
As AI video technology matures, the focus appears to be shifting from viral demos toward sustainable, high-value use cases in productivity, entertainment production, and robotics. Users disappointed by Sora’s exit may find viable alternatives in Google’s Veo tools or other emerging platforms.
For the most current details on access to Veo or timelines for Sora’s full wind-down, refer to official announcements from Google DeepMind or OpenAI. The AI video space remains dynamic, with rapid advancements expected in the months ahead.
About The Author
