The funding model involves private capital trusts that have been forced to limit withdrawals. The big data centre may soon be a half dozen linked machines or less. Because no one will want their IP in data centres they do not own and control.
Pretty much my thoughts. It was a huge loss leader like YouTube, but without a huge company with other sources of income to subsidize it. It's not a failure of AI, but of financial forethought.
The $200 Pro subscription gave users 100 generations per day, and that's in addition to all the ChatGPT and Codex benefits. It's safe to say OpenAI was heavily subsidizing the cost in hopes it might turn into opportunities down the road. No other video generation AI offers that kind of value.
I'm in some groups with all the top creators from Sora, and we've known for some time the writing was on the wall. Sora has had serious user retention problems from the start. Basically they check out the platform for a day or two and leave.
No one I know even knew about the social network aspect to Sora until they checked it out. Perhaps if the execution and marketing had been better things would be different, but I'm doubtful given the points mentioned in this article.
It seems like Sora was never really a priority for OpenAI. They ran it like a startup with a skeleton crew. Sam Altman wasn't engaged in it at all either. He posted a handful of videos when it launched with simple prompts. Conversely, I've heard Zuckerberg uses all the Meta products every day and keeps a close eye on them.
if $200 is their top offering, it's almost certainly worth it to them. they'd have been betting on those users covering the costs created by the others.
obviously, they lost that bet, but the technology is not going away. someone else will figure out a way of making it work if they don't.
The system self-repairs from the exponential, high-entropy, top-down forced path into a bottom-up constructive approach with consideration for resources, energy, environment and people.
The Gartner Group's Tech Hype Cycle has remained undefeated for decades. I do not expect that to change now.
https://en.wikipedia.org/wiki/Gartner_hype_cycle
The funding model involves private capital trusts that have been forced to limit withdrawals. The big data centre may soon be a half dozen linked machines or less. Because no one will want their IP in data centres they do not own and control.
Pretty much my thoughts. It was a huge loss leader like YouTube, but without a huge company with other sources of income to subsidize it. It's not a failure of AI, but of financial forethought.
The $200 Pro subscription gave users 100 generations per day, and that's in addition to all the ChatGPT and Codex benefits. It's safe to say OpenAI was heavily subsidizing the cost in hopes it might turn into opportunities down the road. No other video generation AI offers that kind of value.
I'm in some groups with all the top creators from Sora, and we've known for some time the writing was on the wall. Sora has had serious user retention problems from the start. Basically they check out the platform for a day or two and leave.
No one I know even knew about the social network aspect to Sora until they checked it out. Perhaps if the execution and marketing had been better things would be different, but I'm doubtful given the points mentioned in this article.
It seems like Sora was never really a priority for OpenAI. They ran it like a startup with a skeleton crew. Sam Altman wasn't engaged in it at all either. He posted a handful of videos when it launched with simple prompts. Conversely, I've heard Zuckerberg uses all the Meta products every day and keeps a close eye on them.
It does not look like that a 200$ subscription would be enough.
A user can easily burn through tokens worth ten, hundreds or even thousand times more.
These kind of workloads are expensive and dont scale well. There might be a sweet spot somewhere but its not 200$ a month, far from it.
Usage based pricing would solve a lot of issues for the AI industry but that will be a tough sell.
if $200 is their top offering, it's almost certainly worth it to them. they'd have been betting on those users covering the costs created by the others.
obviously, they lost that bet, but the technology is not going away. someone else will figure out a way of making it work if they don't.
The system self-repairs from the exponential, high-entropy, top-down forced path into a bottom-up constructive approach with consideration for resources, energy, environment and people.
We avoided a black hole.