OpenAI Pulls The Plug On Sora: What This Means for AI Video?

OpenAI’s experiment to turn AI video generation into a social product has come to an unexpectedly swift end.

In a brief post on X, the company confirmed it was shutting down the standalone Sora app, thanking users who created with Sora, shared it, and built a community around it, while promising further details on timelines for the app, API access, and preservation of user creations.

On Instagram, the company mentioned that the decision was part of managing broader research priorities.

An OpenAI spokesperson told Engadget that the company has “decided to discontinue Sora in the consumer app and API”, which reportedly affirms that the API access will be sunsetting soon after.

What Just Happened

The announcement marks a striking reversal for what was widely seen as one of the most ambitious attempts to mainstream generative video. Especially, right after a week, when the company recently introduced an editing feature for Sora on iOS and the web.

Some background: When OpenAI first unveiled Sora, and later its upgraded Sora 2 model, it positioned the system as a major leap toward world simulation capable of generating physically accurate scenes, synchronised dialogue, and cinematic visuals.

The original Sora reveal in early 2024 was described internally as video’s “GPT-1 moment,” validating that scaling neural networks on large video datasets could produce emergent behaviours like object permanence and realistic motion. 

By late 2025, OpenAI had gone a step further, launching a social-first iOS app that allowed users to create and remix videos, even inserting realistic avatars of themselves into AI-generated scenes.

According to reporting by TechCrunch, the Sora app peaked at roughly 3.3 Mn downloads in November, before sliding to about 1.1 Mn downloads by February, a stark contrast to the massive scale of ChatGPT, which has reached nearly 900 Mn weekly active users. Over its lifetime, the app is estimated to have generated just $2.1 Mn in in-app purchase revenue, primarily from users buying additional generation credits.

For a company already burning billions on AI infrastructure, the revenue-to-compute equation may simply have stopped making sense.

The shutdown also comes against the backdrop of mounting safety and legal challenges. Despite OpenAI’s guardrails, users found ways to generate deepfakes of public figures who had not opted in. 

AI-generated videos featuring figures such as the late civil rights leader Martin Luther King Jr. and the late actor Robin Williams circulated online, prompting their estate and family members to publicly urge users to stop generating their likeness through such AI platforms.

Besides this, AI-generated content featuring copyrighted characters — from Mario to Pikachu — further raised the spectre of IP disputes. 

In what could have been a landmark alliance, Disney had reportedly explored a licensing deal and even floated a $1 Bn investment tied to Sora’s future. But with the app now discontinued, that partnership appears to have collapsed before any funds changed hands.

While Sora 1 was discontinued earlier this month, on March 13, the core model, Sora 2 continues to exist in more controlled environments, including paid access tiers and API pathways. It is unclear when the API access will be revoked. 

However, the decision to sunset the social product suggests OpenAI may be reassessing where generative video fits within its broader AI roadmap — particularly as reasoning models and enterprise applications command increasing focus.

The development also comes at a time when AI video creation is becoming faster, cheaper and dramatically more accessible. From solo creators producing episodic content to studios experimenting with AI-led film workflows, output across formats is scaling rapidly. As production barriers fall, industry observers say long-term value may increasingly shift towards IP ownership, distribution control and monetisation leverage rather than sheer content generation.

The bigger question now is what this means for the broader AI video ecosystem.

Sora Shutdown: A Blow For AI Video?

For startups building in AI video, the Sora shutdown is less a shock than a reminder of the category’s structural challenges.

Mandar Natekar, cofounder and CEO of NeuralGarage, said, “Text-to-video is phenomenally high compute cost. It bleeds. It requires massive amounts of GPUs, and there is not enough revenue to justify it. Among AI verticals, generative video has the worst revenue versus resources ratio.”

NeuralGarage focuses on enhancing and transforming existing production footage rather than generating entire scenes from prompts. This distinction, Natekar argues, insulates the company from volatility in the text-to-video market.

However, he acknowledged that the shutdown reflects deeper economic realities.

In that sense, Natekar believes OpenAI’s decision to redirect resources toward reasoning models may be pragmatic rather than alarming.

Ritwika Choudhury, founder and CEO of Unscript AI, echoed a similar stance and said that the company operates at a fundamentally different layer in the AI video stack, insulating it from volatility at the foundation model level.

“We are an AI video engine that enables production houses, microdrama platforms and OTT teams to create full-length films with character consistency, stylistic continuity and directorial control across hundreds of minutes of runtime,” she told Inc42.

According to Choudhury, Unscript has always followed a multi-model architecture, running specialised models across different stages of the video pipeline alongside proprietary systems developed through years of in-house research.

“For simpler use cases, Sora’s exit may push more companies towards multi-model thinking. But at the quality end of the market, this was never optional,” she added.

Choudhury also cautioned against reading too much into the shutdown as a signal of weakness in the category itself.

“Sora shutting down isn’t a sign that AI video is broken. It’s a sign that building for consumers was the wrong bet,” she said. “These models burn enormous computing power, letting people generate novelty clips. The real business opportunity lies in professional use cases — studios, platforms and enterprise teams where the output has measurable value.”

Her comments point to a broader shift in the AI video ecosystem, where startups are increasingly differentiating through orchestration, workflow intelligence and cinematic-grade control rather than relying on any single foundation model.

The shutdown has also triggered conversations across the global open-source community about what happens next to the technology itself. 

Clem Delangue, CEO and cofounder of Hugging Face, wrote on X, “[It] would be so cool if OpenAI open-sourced Sora as they’re shutting down the app,” adding that such a move would be an impactful contribution to the field and would make the efforts of teams working on it more meaningful.

Ultimately, Sora’s shutdown underscores a broader truth about the generative AI boom: technical breakthroughs do not always translate into sustainable products.

AI video models are improving at breakneck speed. But until infrastructure costs fall or revenue models mature, building large-scale consumer platforms around them may remain a high-risk bet.

[Edited by Nikhil Subramaniam]

The post OpenAI Pulls The Plug On Sora: What This Means for AI Video? appeared first on Inc42 Media.