"Whether Sora lasts or not, however, is somewhat beside the point. What catches my attention most is that OpenAI released this app in the first place.
It wasn’t that long ago that Sam Altman was still comparing the release of GPT-5 to the testing of the first atomic bomb , and many commentators took Dario Amodei at his word when he proclaimed 50% of white collar jobs might soon be automated by LLM-based tools."
That's the thing, this has all been predicated on the notion that AGI is next. That's what the money is chasing, why it's sucked in astronomical investments. It's cool, but that's not why Nvidia is a multi trillion dollar company. It's that value because it was promised to be the brainpower behind AGI.
Was a mistake from the beginning to use language as the basis for tokens and embedded spaces between them to generate semantics. It wasn't thought out, it was a snowball trial and error that went out of control.
The fact that OpenAI is pushing Sora, and Altman now even hinting at introducing "erotic roleplay"[0] makes it obvious: openAI has stopped being a real AI research lab. Now, they’re just another desperate player in a no-moat market, scrambling to become the primary platform of this hype era and imprison users onto their platform, just like Microsoft and Facebook did before in the PC and social era.
I got the feeling when this was released that it was just another metric to justify further investment, they were guaranteed to have a lot of users, they can turn around and say "well we have 2 huge applications and were just getting started" investors don't care too much about product quality we've seen, just large numbers.
OpenAI is making a wild number of product plays at once, trying to leverage the value of the frontier model, brand value, and massive number of eyeballs they own. Sora is just one of many. Some will fail and maybe some will succeed.
It seems true that no company has used frontier models to create a product with business value commensurate with the cost it takes to train and run them. That what OpenAI is trying to do with Sora, and with Codex, Apps, "Agent" flows, etc. I don't think there's more to read into it than that.
The massive cost of this product is unique though (not even counting the copyright lawsuits/settlements coming). I can't think of any side projects that require this level of investment.
I love Cal. I really, really take his thoughts on many things to heart for well over a decade now.
I think he's being a bit harsh here. And there are some confounding factors why.
Yes, we have an AI bubble. Yes there's been a ton of hype that can't be met with reality in the short term. That's normal for large changes (and this is a large technological change). OpenAI may have some rough days ahead of it soon, but just like the internet, there's still a lot of signal here and a lot of work to still be done. Going through Suna+Sora videos just last night was still absoutely magical. There's still so much here.
But, OpenAI is also becoming, to use a Ben Thompson term, an aggregator. If it's where you go to solve many problems, advertising and more is a natural fit. It's not certain who comes out on top of the space (or if it can be shared), but there are huge rewards coming in future years, even after a bubble has popped.
Cal is having a very strong reaction here. I value it, but I wish it was more nuanced.
I could see Sora having a significant negative impact on short form video products like TikTok if they don’t quickly and accurately find a way to categorize its use. A steady stream of AI generated video content hurts the value prop of short form video in more than one way… It quickly desensitizes you and takes the surprise out that drives consumption of a lot of content. It also of course leaves you feeling like you can’t trust anything you see.
> A company that still believes that its technology was imminently going to run large swathes of the economy, and would be so powerful as to reconfigure our experience of the world as we know it, wouldn’t be seeking to make a quick buck selling ads against deep fake videos of historical figures wrestling.
But also, a company that earnestly believes that it's about to disrupt most labor is going to want to grab as many of those bucks as possible before people no longer have income.
This take feels like classic Cal Newport pattern-matching: something looks vaguely "consumerish," so it must signal decline. It's a huge overreach.
Whether OpenAI becomes a truly massive, world-defining company is an open question, but it's not going to be decided by Sora. Treating a research-facing video generator as if it's OpenAI's attempt at the next TikTok is just missing the forest for the trees. Sora isn't a product bet, it's a technology demo or a testbed for video and image modeling. They threw a basic interface on top so people could actually use it. If they shut that interface down tomorrow, it wouldn't change a thing about the underlying progress in generative modeling.
You can argue that OpenAI lacks focus, or that they waste energy on these experiments. That's a reasonable discussion. But calling it "the beginning of the end" because of one side project is just unserious. Tech companies at the frontier run hundreds of little prototypes like this... most get abandoned, and that's fine.
The real question about OpenAI's future has nothing to do with Sora. It's whether large language and multimodal models eventually become a zero-margin commodity. If that happens, OpenAI's valuation problem isn't about branding or app strategy, it's about economics. Can they build a moat beyond "we have the biggest model"? Because that won't hold once opensource and fine-tuned domain models catch up.
So sure, Sora might be a distraction. But pretending that a minor interface launch is some great unraveling of OpenAI's trajectory is just lazy narrative-hunting.
There are also interesting things one could do with models like Sora, depending how it actually performs in practice: prompting to segment, for example; and the thing could very possibly, if it's fast enough etc. become a foundation for robotics.
It seems they are going to try to maximize their installed base, build the infrastructure, and try to own everything in between, whether it’s LLM or some other architecture that arises. Owning data centers and an installed base sounds great in theory, but it assumes you can outbuild hyperscalers on infrastructure and that your users will stick around. Data centers are a low margin grind and the installed base in AI isn’t locked in like iPhones. Apple and Google still control the endpoints, and I think they’ll ultimately decide who wins by what they integrate at the OS level.
Their first bet was than they were going to be the frontier model provider by a good margin, and that others would not be able to compete on the "intelligence". And that they could get distribution via big customers looking to buy model access.
The dominant-model-provider strategy has already failed, many actors have models that rival them - both established (Google) and newcomers (Anthropic). Open models are not to shabby either, enough to undermine the narrative "we are uniquely able to do powerful models". As you say, there is a commodification process started, and it might be a race to the bottom in terms of margin.
So, OpenAI has moved into a new/adapted strategy, where they want to own the customers to a much larger degree, and rely less on partners/customers for distribution. This is likely because their prospective partners have a bunch of viable models to select between (many end products for power users lets people select freely), and high competitive pressure on costs (as it defines the margin and competitiveness) of the end products. Codex, Sora, their new web browser announcement, adjustments in ChatGPT is all to ensure a lot of direct end users - more brand recognition, more influence, more monetization possibilities.
So I think it is a considerable pivot from their initial plan/hopes. But it is not an unraveling - it is a rather smart response to the fierce competition in the market.
> It’s unclear whether this app will last. One major issue is the back-end expense of producing these videos. For now, OpenAI requires a paid ChatGPT Plus account to generate your own content. At the $20 tier, you can pump out up to 50 low-resolution videos per month. For a whopping $200 a month, you can generate more videos at higher resolutions. None of this compares favorably to competitors like TikTok, which are exponentially cheaper to operate and can therefore not only remain truly free for all users, but actually pay their creators .
fwiw, there's no requirement to have a subscription to create content.
This article makes the claim that OpenAI, and AI in general, is massively overhyped because OpenAI is looking to sell slop. I'm not sure I can agree with that basic premise.
Whether AGI does or does not materialize sometime soon doesn't matter. OpenAI, like every company who wants to raise massive amounts of money, needs to show huge growth numbers now. It seems like the unfortunate, simple truth is that slop is a growth hack.
OpenAI is steering significant amounts of traffic away from Google and ChatGPT is a fairly common name that extends beyond awareness of the company (or even what the word it means specifically).
Not nearly on the level of "Kleenex" or "Google" as a term, but impressive given that other companies have spent decades trying to make a similar dent.
When you have this much money, you can afford to chase AGI, and assign a few people to make an app. The app might be frivolous, but it keep OpenAI in the public view. It's money well spent on marketing.
"It wasn't that long ago that Sam Altman was still comparing the release of GPT-5 to the testing of the first atomic bomb, and many commentators took Dario Amodei at his word when he proclaimed* 50% of white collar jobs might soon be automated* by LLM-based tools.
A company that still believes that its technology was imminently going to run large swathes of the economy, and would be so powerful as to reconfigure our experience of the world as we know it, wouldn't be seeking to make a quick buck selling ads against deep fake videos of historical figures wrestling. They also wouldn't be entertaining the idea, as Altman did last week, that they might soon start offering an age-gated version of ChatGPT so that adults could enjoy"
They might be forced to do so because the current inference pricing is not really covered by the 20$ monthly fee.
Who knows what they have promised to the investors and the real cashflow is hard to be certain about with the circular nature of cross-investing between the biggest players.
More likely it's the beginning of the end for TikTok, since the amount of posts that use seems to be flooding the platform, lowering trust and credibility in each video.
Hmm I have my doubts. I don't really understand the appeal of hyper-consumerism, brand focus, and big SM platform engagement. But I look over at my wife and she is clearly plugged in to a giant worldwide cultural movement of womens interests, product recommendations, and politics. It also appears to be the case that women are less likely to engage with AI. It's one of the big reasons I think it's a massive blunder to pursue AI erotica. I think a big part of the TikTok user base (women) will have to be persuaded to jump on the AI bandwagon and I'm not sure the industry is pursuing any products or features to woo that market.
> wouldn’t be seeking to make a quick buck selling ads against deep fake videos
This isn't a money making venture for them, and you basically admitted as much. They poured no doubt massive amounts of money into developing this and have little hope of earning it back soon. This is an attempt to keep up with other ai companies also developing video models in order to not look behind to investors. Making it available to users is similarly about increasing active user counts in order to look more successful. If people incidentally get off to it that's not their concern
The counter argument is that OpenAI has to make fairly bolder moves.
Social was _already_ becoming the domain of AI generated content. In the benign sense, there's been social content of people sharing their silly AI content since early DALL-e. Its a good idea to make a social app that's actually _about that_, because you can remix and play with the content in a novel way.
The first Sora was sort of already going in this direction.
When fast take-off starts to be evident later next year, it’s the fact that OpenAI has built its hoard on a diverse set of product lines with a broad surface area to operate on that will be the differentiator allowing them to lead on the exponential before anyone else.
36 comments
[ 6.0 ms ] story [ 90.9 ms ] threadIt wasn’t that long ago that Sam Altman was still comparing the release of GPT-5 to the testing of the first atomic bomb , and many commentators took Dario Amodei at his word when he proclaimed 50% of white collar jobs might soon be automated by LLM-based tools."
That's the thing, this has all been predicated on the notion that AGI is next. That's what the money is chasing, why it's sucked in astronomical investments. It's cool, but that's not why Nvidia is a multi trillion dollar company. It's that value because it was promised to be the brainpower behind AGI.
People will be swayed by AI-generated videos while also being convinced real videos are AI.
I'm kinda terrified of the future of politics.
[0] https://www.404media.co/openai-sam-altman-interview-chatgpt-...
It seems true that no company has used frontier models to create a product with business value commensurate with the cost it takes to train and run them. That what OpenAI is trying to do with Sora, and with Codex, Apps, "Agent" flows, etc. I don't think there's more to read into it than that.
I think he's being a bit harsh here. And there are some confounding factors why.
Yes, we have an AI bubble. Yes there's been a ton of hype that can't be met with reality in the short term. That's normal for large changes (and this is a large technological change). OpenAI may have some rough days ahead of it soon, but just like the internet, there's still a lot of signal here and a lot of work to still be done. Going through Suna+Sora videos just last night was still absoutely magical. There's still so much here.
But, OpenAI is also becoming, to use a Ben Thompson term, an aggregator. If it's where you go to solve many problems, advertising and more is a natural fit. It's not certain who comes out on top of the space (or if it can be shared), but there are huge rewards coming in future years, even after a bubble has popped.
Cal is having a very strong reaction here. I value it, but I wish it was more nuanced.
But also, a company that earnestly believes that it's about to disrupt most labor is going to want to grab as many of those bucks as possible before people no longer have income.
Whether OpenAI becomes a truly massive, world-defining company is an open question, but it's not going to be decided by Sora. Treating a research-facing video generator as if it's OpenAI's attempt at the next TikTok is just missing the forest for the trees. Sora isn't a product bet, it's a technology demo or a testbed for video and image modeling. They threw a basic interface on top so people could actually use it. If they shut that interface down tomorrow, it wouldn't change a thing about the underlying progress in generative modeling.
You can argue that OpenAI lacks focus, or that they waste energy on these experiments. That's a reasonable discussion. But calling it "the beginning of the end" because of one side project is just unserious. Tech companies at the frontier run hundreds of little prototypes like this... most get abandoned, and that's fine.
The real question about OpenAI's future has nothing to do with Sora. It's whether large language and multimodal models eventually become a zero-margin commodity. If that happens, OpenAI's valuation problem isn't about branding or app strategy, it's about economics. Can they build a moat beyond "we have the biggest model"? Because that won't hold once opensource and fine-tuned domain models catch up.
So sure, Sora might be a distraction. But pretending that a minor interface launch is some great unraveling of OpenAI's trajectory is just lazy narrative-hunting.
ChatGPT clearly is "for consumers". Whereas Sora is a kind of enshitification to monetize engagement. It's right to question the latter.
fwiw, there's no requirement to have a subscription to create content.
Whether AGI does or does not materialize sometime soon doesn't matter. OpenAI, like every company who wants to raise massive amounts of money, needs to show huge growth numbers now. It seems like the unfortunate, simple truth is that slop is a growth hack.
Not nearly on the level of "Kleenex" or "Google" as a term, but impressive given that other companies have spent decades trying to make a similar dent.
A company that still believes that its technology was imminently going to run large swathes of the economy, and would be so powerful as to reconfigure our experience of the world as we know it, wouldn't be seeking to make a quick buck selling ads against deep fake videos of historical figures wrestling. They also wouldn't be entertaining the idea, as Altman did last week, that they might soon start offering an age-gated version of ChatGPT so that adults could enjoy"
This isn't a money making venture for them, and you basically admitted as much. They poured no doubt massive amounts of money into developing this and have little hope of earning it back soon. This is an attempt to keep up with other ai companies also developing video models in order to not look behind to investors. Making it available to users is similarly about increasing active user counts in order to look more successful. If people incidentally get off to it that's not their concern
Social was _already_ becoming the domain of AI generated content. In the benign sense, there's been social content of people sharing their silly AI content since early DALL-e. Its a good idea to make a social app that's actually _about that_, because you can remix and play with the content in a novel way.
The first Sora was sort of already going in this direction.