Blog

This week, OpenAI revealed its decision to deactivate its Sora application and associated video models, merely half a year after the app’s initial launch.

During the most recent installment of TechCrunch’s Equity podcast, Kirsten Korosec, Sean O’Kane, and I explored the implications of this action for OpenAI and the broader industry. To some degree, this move appears consistent with previous indications that OpenAI is prioritizing enterprise and productivity tools, especially in anticipation of a potential IPO.

In fact, Kirsten proposed that OpenAI’s choice to discontinue Sora represented “a welcome sign of maturity from an AI research lab.”

However, Sora’s termination — alongside ByteDance’s reported deferral of Seedance 2.0’s global release, their video generation model — might also serve as a reality check for creators of AI video technologies and enthusiasts who swiftly declared these tools would soon replace Hollywood.

Below, you can find an edited excerpt of our discussion, condensed for clarity and length.

Anthony: I believe it’s important to emphasize that this isn’t solely about the application. The app, to me, and I think to others, was particularly unappealing due to its concept of a social network devoid of people, filled with nothing but ‘slop’.

Yet, beyond the app, it appears OpenAI is effectively scaling back nearly all its video-related initiatives. As reported by the Wall Street Journal, which initially broke parts of this story, the core reason is that OpenAI — potentially in preparation for going public — is intensely focusing on business, enterprise, and programming products. Consequently, this consumer social app, and video more broadly, is currently not a strategic priority.

Sean: Indeed, I never truly engaged with [the app]. The very idea of it deterred me for several reasons. And you know, it served as a valuable reminder that OpenAI — and I say this without intending to diminish their achievements in any way — but I think this likely reminded them internally of the element of fortune […] involved in ChatGPT’s remarkable success.

Clearly, there is genuine value in it for users; I don’t wish to discredit that, because you don’t achieve the usage figures we’ve heard without something fundamentally working right — and even more so, without it being sustained over years and evolving into something that remains relevant to people.

However, with Sora’s launch, there was a sense of, “We built the most successful consumer product ever, and now we’re doing it again. And we’re going to bring in Disney and all this stuff.” I believe this is just a very stark reminder that it won’t always be an automatic path to becoming the next top consumer product, and that there truly needs to be something users find meaningful for it to endure.

Kirsten: Yes, I actually commend OpenAI for this decision. While we sometimes satirize the “move fast and break things” philosophy, I do think there’s merit in companies that can innovate rapidly and then discontinue products that aren’t performing, without viewing it as a failure. Of course, real money was expended. If one considers the deal with Disney, that was a billion-dollar agreement, but if you look at — and we lack the insight into this since we don’t examine their balance sheets — what exactly they were investing in this and its long-term value for the corporation?

And I believe that while, certainly, it was fascinating to observe their creations, their choice to close it, to me, demonstrated a level of maturity that was reassuring to witness in an AI lab.

Anthony: Regarding its implications for OpenAI, this seems entirely consistent with everything we’ve been learning about their future strategy. It doesn’t appear to be a significant setback or anything of that nature when considering the future of generative AI.

In the context of video specifically, it’s intriguing because this also coincides with reports surrounding Seedance, which is ByteDance’s generative AI model for video. There are indications that [Seedance 2.0 has] been postponed due to engineering and legal inquiries, essentially grappling with, “Can we embed IP protections into this?” Which, apparently, they hadn’t previously addressed with sufficient seriousness.

Thus, it’s a moment of reckoning. There were highly exaggerated assertions, even from individuals within Hollywood, suggesting something like, “We’re finished, this is the future; it’s just about typing prompts and producing feature films.” And it turns out that, for a myriad of technical and legal reasons, it isn’t nearly that straightforward, and we are still very, very far from that scenario unfolding.

Sean: And the final point I think we should make about this is that it’s one of several decisions that appear to be occurring after Fidji Simo assumed her role, essentially overseeing daily operations. That constitutes a massive shift in dynamics within OpenAI. And I believe the further we move from the moment she was appointed to run the show, especially regarding these consumer products and determining their fate, the easier it will be to reflect on this period and recognize its significance for the company.