OpenAI reportedly pulls plug on large video model Sora as revenues fail to cover soaring costs

What happened

It has been reported that OpenAI abruptly shut down its large video model, Sora, after a short run in public testing. The move reportedly followed internal assessments showing that revenues from Sora were far below the enormous operational costs required to run and scale the system. Details remain sparse and OpenAI has not publicly released a full accounting, but multiple industry sources say the shutdown was sudden.

Why it stalled

Sora was positioned as a next‑generation video generation and editing model — a natural extension of text and image models that promises to automate complex video production tasks. But generating high‑quality video at scale demands huge GPU clusters and expensive inference infrastructure. Reportedly, user uptake and monetization did not grow quickly enough to offset those compute bills. Was this a technology problem or a business‑model one? Experts argue it was both: the tech is expensive, and pricing consumer or creator use at a level that covers costs risks throttling demand.

Bigger picture and geopolitics

The setback highlights a wider industry reality. Advanced AI workloads are tightly coupled to scarce, costly chips and cloud capacity. Geopolitical factors — including U.S. export controls on advanced accelerators and tighter trade restrictions — have constrained supply and driven costs higher, complicating margins for emerging products. Competition is also fierce, with major cloud and AI players testing similar offerings; that makes rapid monetization even harder.

What this means going forward

For Western users and investors watching the AI arms race, the Sora episode is a reminder that breakthrough capabilities do not guarantee sustainable businesses. Companies will need clearer paths to profitability, cheaper inference, or novel monetization approaches for heavy‑compute models. It has been reported that some teams may pivot the underlying research into more narrowly focused enterprise or studio tools — where customers can better absorb the cost — rather than broad consumer services.