Why is OpenAI like this?

If messy drama was something you could sell, OpenAI would have turned a profit already.

That much is clear from the bizarre way the ChatGPT company killed off Sora, its shortform video app — just one of multiple dramas that seem to be playing out in tandem at Sam Altman's company. So much so that you'd be forgiven for wondering whether there is any method to this madness.

The Sora shuttering could not have burned more bridges with partners and creators if it tried. The rumored face-saving announcement that Sora was being folded into ChatGPT? Didn't happen. OpenAI had just announced new Sora safety standards the day before — so why bother if it's on the outs?

Disney, meanwhile, had not yet finalized its $1 billion, three-year deal with OpenAI over character licensing in Sora. The House of Mouse was so blindsided, according to a Reuters report, that executives had a meeting discussing what Disney could do with the Sora deal yesterday... just 30 minutes before OpenAI announced its death via a post on X.

ChatGPT ads are a bust

There are also unforced errors in OpenAI's ad division, according to a scoop from The Information. It seems an early test of ChatGPT ads (the company's big revenue-hunting pivot from two months ago) burned bridges with the advertisers who participated. Those advertisers bought more than $200,000 of ads apiece, premium table stakes for the industry. But they didn't get the data they need to know if their ads — which played minimally — are paying off.

Ironically, considering how much OpenAI's products promise to automate the world, the ad industry titans said they had to call or email the company to place ads.

And where is OpenAI's CEO in all this? Not tweeting through it; at time of writing, Altman hasn't posted on X since the Sora announcement. Nor is he blogging through it; his last two blog posts were written six months ago, and are both about Sora. "People are generating much more [video] than we expected per user," Altman wrote in his most recent post.

"We are going to have to somehow make money for video generation," the CEO said, noting presciently: "please expect a very high rate of change from us."

No kidding. So where is the money going to come from, if it isn't coming from Disney and it isn't coming from ads? Shuttering Sora won't make up OpenAI's revenue shortfall, estimated at roughly $1 billion a month. Running ChatGPT for 900 million people per week, all but 40 million of whom aren't paying for it, costs a lot!

Nor is the money coming from the Instant Checkout shopping feature OpenAI announced in 2025, which was supposed to attract a million merchants. The company announced Tuesday that Altman is dumping that as well.

OpenAI's race to its IPO

So what is Altman doing, exactly? He told OpenAI staff on Tuesday that he's no longer running the company's safety and security teams, according to a separate Information report. Instead, Altman reportedly told them that he's focused on raising more money and building more data centers.

"Things are moving faster than many of us expected," Altman added, though in which direction they're moving wasn't clear.

Altman, according to the report, is renaming the company's product group "AGI deployment." A ballsy marketing move, given that GPT-5 is nowhere near AGI (Artificial General Intelligence, also known as digital superintelligence) — and OpenAI is still feuding with its partner Microsoft about how we would define such a thing if it did arrive.

That's no mere academic debate, given that Amazon reportedly stands to invest another $50 billion in the company on the condition that it goes public, or AGI is achieved.

In the absence of actual AGI to deploy — and there are no signs that OpenAI's next model, codenamed "Spud," will get there either — there's nothing left for Altman to do but prepare for the IPO. OpenAI and its chief rival Anthropic, which makes Claude, are both supposedly on course for their public offerings this year. As public companies, they'll be able to raise funds more easily.

But what if there's only room for one winner?

Anthropic, while it's still (sort of) battling the Pentagon and coming up with its own creative revenue estimates, is widely seen as the more successful company when it comes to solid, dependable business. Not for them the flashy Disney deals; they're selling targeted AI services to other companies. Right now, the perception among customers is that Claude is for coding and making AI agents, while ChatGPT is more for the masses.

It isn't going to be easy for Altman to turn that perception around. But it'll probably be easier if OpenAI isn't trying to be a media company that sells ads and lets you make cute Disney character shorts. And in that context, Altman's latest moves start to make a lick of sense. Whether he can convince business customers that he's a dependable figure — pay no attention to those burning bridges! — will be another drama altogether.