OpenAI Killed Its Most Talked-About App Six Months After Launch. Here's Why That Should Worry You
When Sora launched, it felt like the app everyone had to try at least once, even if just to see their face turned into something ridiculous. It topped the app store charts within days. People I know who'd never touched an AI tool in their life were suddenly making short clips of themselves doing things that never happened. Six months later, it's gone. Not scaled back, not merged into something else. Shut down entirely.
That's a strange enough turn on its own. What's more useful, though, is understanding why, because the reason says a lot about where AI tools are actually heading, and it's not the story most people assume.
It wasn't about the technology failing
The easy assumption is that Sora got shut down because it stopped working well, or because a competitor did it better. Neither is really true. The reporting on this, and it's been reported in real detail, points to something far more boring and far more telling: money. Sora's operating costs reportedly ran around a million dollars a day. Over its entire lifetime, in-app purchases brought in about two point one million dollars total. Read that again if you need to. A million dollars a day going out, two million total coming in across the whole run.
Generating video with AI is enormously expensive compared to generating text. Every clip burns through serious computing power, and OpenAI apparently decided that pouring that capacity into a flashy consumer app wasn't worth it when the same resources could go into the coding and enterprise tools that are actually bringing in money ahead of a planned public listing. There was also a Disney partnership reportedly worth close to a billion dollars that fell apart around the same time, which didn't help the case for keeping it alive.
There's a quieter lesson buried in there for anyone who builds things online. A product can be genuinely popular, get a million users at launch, generate a ton of buzz and attention, and still be a financial sinkhole underneath. Attention isn't revenue. It never automatically was, and this is about as clean an example as you'll find of that gap.
What this means if you actually make video content
If any part of your work depends on AI-generated video, whether that's short clips for social media, product demos, or ad creative, the practical question isn't "what happened to Sora," it's "what do I use now." And the honest answer in mid-2026 is: probably more than one tool, not one replacement.
Google's Veo 3.1 has become the default answer for a lot of people who want the highest quality output, partly because it generates synced audio in the same pass as the video, meaning you're not stitching sound on afterward, and it can output true 4K. Kling 3.0 has carved out its own space for anything involving multiple shots that need to feel connected, since it can hold a shared audio timeline across a sequence and handle native dialogue in several languages, which matters if you're making anything that resembles a short story rather than a single clip. Runway's Gen-4.5 tends to come up specifically when editing matters as much as the initial generation, for people who need to actually shape and cut the output rather than just generate and post it as-is.
What surprised me researching this is that almost nobody serious about video seems to be using just one of these. The common pattern is picking one tool for the hero shot, the thing that actually needs to look impressive, and a faster, cheaper tool for everything else you need to churn out quickly. That's a more realistic workflow than the "one app does it all" pitch every one of these platforms leads with in its marketing.
The pattern worth actually remembering
Sora isn't the first AI product to launch huge and quietly disappear, and it won't be the last. What makes it worth writing about isn't the app itself, it's the reminder that the AI tool landscape right now moves fast enough that building your entire workflow around one platform is a genuine risk, not just an inconvenience if that platform disappears.
I've started treating any AI tool I rely on for actual work the way I'd treat a single supplier for my business: useful, maybe even great, but not something I want to be fully dependent on if it vanished tomorrow. Sora had a million users and a billion-dollar partner and still didn't survive its first year in that form. Popularity isn't the same as permanence, and anyone building a content workflow around AI tools right now should keep that in the back of their mind, not out of pessimism, just because it happened, in public, to one of the most talked-about apps of the year.
If there's a practical takeaway beyond "know your backup tools," it's this: the AI products that survive long term are probably going to be the ones solving a problem people will pay for directly, not the ones that are simply fun to try once. Sora was genuinely fun. It just wasn't that.