The decision does not mean the end of Sora itself. The underlying model will continue to be developed, but as part of a broader strategy centred on enterprise products, coding tools and longer-term research into “world models” for robotics.

Sora attracted significant attention for its ability to generate realistic video from text prompts. But it also came with a heavy price: video models are among the most compute-intensive workloads in AI, requiring vast processing power to run at scale.

That appears to have been a key factor behind the shift. As demand for AI infrastructure accelerates, OpenAI is reallocating resources towards areas with clearer revenue potential and lower operational strain.

The move reflects a wider recalibration across the sector. AI leaders are increasingly balancing innovation with the realities of limited compute, rising costs and the need to monetise effectively. In that context, enterprise use cases, where spending is more predictable, are taking priority over viral consumer applications.

There were also external pressures. Sora faced ongoing scrutiny around copyright and deepfake risks, adding regulatory complexity to an already demanding product. Combined with questions over sustained user engagement, the case for a standalone app appears to have weakened.

Instead, OpenAI is doubling down on integrating its models into broader platforms and tools, while advancing research into systems that can better simulate real-world environments, seen as a key step towards robotics and autonomous technologies.

RELATED STORIES

OpenAI secures $1bn investment from Disney in historic licensing deal

Meta unveils Movie Gen: AI video creator to rival OpenAI’s Sora