OpenAI on Tuesday said it will discontinue its once-viral AI video app, Sora, 176 days (or about 6 months) after it was initially released. It bravely asked the question: Do we really need this? For once, it came to the right answer. No, no, we do not.

This is the biggest, most public project OpenAI has killed. While it certainly shows a lack of confidence in the generative media side of things, I don't think it's a sign that the AI industry is collapsing. (Sorry if that's what you were hoping for.) The true story is a bit deeper. 

If OpenAI had wanted to build the best AI video tool or invent a new kind of social media, it could and would have. But Sora is an odd duck. The second-generation model is impressive, nabbing a slot in CNET's ranking of AI video tools. But the social media app is bizarre. Half AI, half social media, all fake. 

Whatever Sora was meant to be, it never lived up to the dream. But there's still a lot to learn from Sora's rapid rise and sudden death.

There's always a small chance OpenAI changes its mind -- just look at Meta, which pulled the plug and resurrected the Metaverse in two days. But I think the company -- and all of us who have to live in this age of AI -- will be better off if it stays the course. Here's why.

Sora was never OpenAI's endgame

Here's the dirty secret of generative media: It's incredibly costly. It takes a lot of developer work to create a model that doesn't spit out embarrassingly bad results. So it's expensive before it's even released. Once it is, it requires a lot of compute to render complex videos and images compared to relatively simple text. On top of that, it's controversial. You'll probably get sued for copyright infringement at some point, but that's nothing new. (Disclosure: Ziff Davis, CNET's parent company, in 2025 filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)

It's not a game you want to be in unless you're all in. OpenAI had two paths it could've gone down with Sora: Full social media or full AI video. It didn't do either. OpenAI was never all in on Sora the way it is with ChatGPT.

The company never invested the resources it needed to turn its video model into a state-of-the-art, professional-grade tool. Sora 2 had impressive audio and visuals, yes. But you couldn't easily edit your videos. Its storyboarding tool never lived up to my expectations, and I'm no professional. Meanwhile, Google built a professional editing program for its AI tools, called Flow, and Adobe incorporated its AI into its existing industry-standard editing programs. Sora 2, the model, is great, but it was limited by where it sat inside the app and website.

So if Sora wasn't going to be a professional tool, at least not without a lot of work, then its primary purpose was making memes. It was weird that OpenAI was willingly getting into the social media business. A newer AI video model, sure, that made sense. But running a social media platform is a hard job that comes with a lot of responsibilities, hard choices and ethical imperatives. Content moderation alone is a juggernaut that should've sunk Meta more than once. (And a New Mexico jury slapped Meta with $375 million in penalties over its moderation and safety failures less than an hour after the Sora news broke Tuesday.) Sam Altman never indicated that he was interested in becoming the AI version of Mark Zuckerberg.

(Meta did actually release an AI video app shortly before OpenAI did, called Vibes. You probably didn't know this because nobody cared about it.)

Sora never developed a true social media personality. There were no major breakthrough cultural moments. It didn't have TikTok dances or Instagram's legacy, for example. There were no brands paying for ads or famous creators drawing in new users. OpenAI was probably losing a lot of money just to keep the app running. We can make memes without AI and, frankly, the last thing we need is more AI slop to scroll through. 

It wasn't until OpenAI struck a $1 billion deal with Disney that it seemed like Sora may have been thrown a lifeline. If OpenAI had not shut down Sora, I would've bet that Disney might have given Sora the boost it needed. The ability to make videos legally that feature recognizable characters would've attracted new users and reignited fans' fervor. But the world isn't really missing out on much without the possibility of more Spider-Man-themed slop.

Instead, pulling the plug on Sora gives OpenAI the chance to pursue that most elusive goal for tech companies: Actually useful AI.

So what is the endgame?

While the latter half of 2025 was all about AI image and video tools, 2026 is all about AI that does stuff, particularly in the workplace: agentic AI, coding agents, claws and robots.

Anthropic set off a race this year when its advanced Claude Code and Cowork tools proved that AI could be a serious workplace tool. OpenAI released Codex to compete but Anthropic's popularity and new user sign-ups have been rising. Fidji Simo, OpenAI's head of applications, reportedly told employees earlier this month that it would be cutting down on "side quests" to focus on more business tools. By comparison, meme AI videos certainly don't seem like a top priority.

Realistically, these business-centric tools are more appealing to customers who may pay OpenAI real money for their AI services. Whatever OpenAI's true financial situation is, I imagine they want that business. It also proves the company is more than just the first to make generative AI mainstream. It can create value beyond error-ridden information, sycophantic pseudo-therapists and the horror that is AI romantic companions.

Everyone loves to talk about the AI industry moving fast, breaking things and asking for forgiveness later, etc. But if one of the things the AI industry kills is the useless stuff it created in previous years, I think we'll all be better off for it.

I hope Sora's death inspires us to have a serious conversation about what we actually need our AI tools to do. Because it isn't AI memes. It isn't a lot of the other things AI companies want you to believe. The first step toward ridding ourselves of useless AI -- in our smartphones, artwork and culture -- is recognizing we don't need it.