Last year, the highly hyped Sora only lasted 7 months before shutting down, which is truly disappointing!

I remember when it launched in September last year, Sora hit over one million downloads in five days and briefly topped the App Store charts. I even recommended it to my friends. But seven months later, the 30-day retention rate was just 1%, and at 60 days, it dropped to zero. I think the reason Sora failed so quickly is partly because it was a pseudo-necessity—users only used it for entertainment emojis, pure visuals for fun, with no repeat purchases or monetization. On top of that, burning money was an endless pit—daily operating costs reached $15 million, and generating 10-second videos cost at least $1.30. Its business model couldn’t keep up, and combined with poor community engagement and user experience, it was impossible to retain users.

However, Sora’s demise isn’t just a product issue; it’s also part of OpenAI’s strategic contraction as it rushes toward an IPO. To impress the capital markets, OpenAI has to cut back on side projects and focus on core initiatives—shutting down Sora and reallocating computing power to the upcoming next-generation main models. They are fully betting on the enterprise market and Agentic AI. Even the Sora team has shifted to researching world models related to robotics.

In the past couple of years, everyone was expanding wildly, trying to jump on every trend—text, video, chat, hardware—fearing missing out on any opportunity. Now, in this AI marathon, running fast isn’t as important as running correctly. Pseudo-demand, high costs, and products without a clear business loop simply can’t go far. Sora isn’t the first to fall, and it won’t be the last.