Well unless they have reach a level where training is not needed (which will be never for an LLM), what you told me is that you make a profit freezing ice cream but lose it on the eggs, cream and sugar necessary to make it
The model releases are getting shorter and shorter. close to a month now. There is no actual profitability not because the price isn’t higher than cost of ingredients, but because they don’t count throwing away the ice cream mixer, and factory every month, when they should if replacing them is part of the process.
Training is not needed to run inference on the existing models, it’s needed to make new models. If LLMs plateau and there’s no use in training new models, or the government regulates them and they can’t train new frontier models, then they can still run inference on the currently existing models and make a profit.
This is more like making a profit for making ice cream but losing a bunch of money researching new recipes for ice cream. If you decide the recipe is good enough and stop doing that research you can still make money making and selling ice cream.
so a model trained in 2 year old data is just as good? as a model trained today?
Yes, assuming a plateau in LLM capability the only reason you’d train on an updated corpus is to get fresh info. That’s not worth it because:
The model will most likely be wrapped in a harness that has search capabilities, so it can use that to fetch fresh info
2 year old data may actually be better as the Internet becomes increasingly tainted with LLM content
Most of the improvement from new models isn’t coming from expanding or updating the corpus, its coming from increasing the parameter size and reinforcement learning.
So they are just choosing to lose money?
No, they are competing. If open AI decides to stop training new models right now then anthropic will and take all there business as the switching cost for models is low. That’s why they want the government to regulate it, so they can have a ceasefire to start taking in profits from inference without worrying about their competitor making a new better model and eating there lunch.
if that were true, and the current models were actually good, there would be wide adoption of the current models while people wait for the next leap… instead, you get glimpses of the terrible numbers like 3% of user base is actually paying for these things
maybe we torture our analogies to death but you make it sound like they have the formula for coca cola but instead of selling that, they are burning cash trying to find an even better formula
when the reality is that they have something that is technically drinkable and not officially poison and they are selling it like the next elixir of life (while they burn cash trying to catch up to their own hype)
People are adopting the current models, mostly coders right now. Most people using LLMs are using them for basic information search, ie. Google AI overviews. Those tasks don’t require the top models and don’t burn that many tokens so the companies keep them free to get the public exposed to AI.
Then there’s the 3% of people who are the power users using it for work, especially coders. For productivity the top models do perform better and the stakes are higher so they need to perform better. The top models aren’t needed for every task, but they shine as an orchestrator of smaller models handling more basic tasks. Coding also requires a lot more tokens, to read all the existing code; to do “thinking” which generates a bunch of output tokens to imitate reasoning; then to generate the code itself; then to review it, adjust after a review…
All of that equates to large bills for token spend to anthropic or open AI. My Claude bill on my company account is approaching $2,000 this month, and that’s about average / what’s expected from an engineer at my company. I’d bet that every engineer in silicon valley is burning through a similar amount as well.
This is why both open AI and anthropic are both seeing extremely high revenue growth. Anthropic ended 2025 with $9b in revenue , they are now on track to hit $100b revenue this year, for reference / the analogy coca cola made $47b this year, it’s a larger revenue then every other software company except Microsoft. THAT IS INSANE, no company has ever seen that kind of growth. Yes it’s being weighed down by training costs so they aren’t profitable but if they even double there revenue next year that could change.
People are adopting the current models, mostly coders right now.
Models are being push down people’s throats, mostly to coders right now… there FIFY
Most people using LLMs are using them for basic information search, ie. Google AI overviews. Those tasks don’t require the top models and don’t burn that many tokens so the companies keep them free to get the public exposed to AI.
Again, this is forced, you literally cannot perform a google search (or bing) without getting some AI results whether you want it or not. Then AI companies, like you, use this to pretend it’s actual demand for their products
… All of that equates to large bills for token spend to anthropic or open AI
If 3% of people generate the cost the AI companies cannot keep to the prices they charge, then this is the definition of an unprofitable business
This is why both open AI and anthropic are both seeing extremely high revenue growth. Anthropic ended 2025 with $9b in revenue , they are now on track to hit $100b revenue this year
No, this is what they ARE CLAIMING THEY WILL REACH. Last year, Altman said it was ridiculous that people were saying they generated $12 billion in revenue… just to then show it was less than that. Now they are delaying their IPO which means the books stay closed for another while longer
THAT IS INSANE, no software company has ever seen that kind of growth
Again you are just drinking the koolaid here (not counting all the anecdotal evidence you now apply to all of Silicon Valley). The truth is that whatever revenue they actually have (or claim, “trust me bro”), you can’t even know if half of it comes from new investments, creative accounting or selling off assets, you just assume it is more people demanding more of their services and paying through the nose for it even if they have barely showed any tangible results in real life https://fortune.com/2026/08/22/executives-ai-productivity-layoffs-study/
but if they even double there revenue next year that could change.
Oh that’s it? just double the already inflated revenue they are reporting without any evidence? of course!!! and all I need to fly is to grow some wings, that’s it, it’s really that simple
I don’t think you understand the scale, even if half of that number is real that’s $50b , that’s larger then coca cola, that’s larger then Salesforce and every other software company except Microsoft. That’s 5x growth on $10b , that is unheard of in the software industry, much less any other industry.
IDK what to tell you if you don’t believe the numbers, yes they aren’t official but every credible source is pointing to massive revenue growth. We’ll see the actual numbers in the coming weeks as they prepare for the IPO, and if they do turn out to be cooking the books to the tune of $50b , I’ll come back to this comment and admit I’m a credulous idiot who fell for the hype. But if they are true I’ll also be coming back with an I told you so.
Also do you think they would cook the books to that extent if they are still going ahead with an IPO? Fabricating $50b in revenue would hurt an IPO once it’s found out way more then some temporary hype from a NYT article.
that’s larger then coca cola, that’s larger then Salesforce and every other software company except Microsoft. That’s 5x growth on $10b , that is unheard of in the software industry, much less any other industry.
and you only need the to double it (the full 100 billions} to make a profit lol…
IDK what to tell you if you don’t believe the numbers, yes they aren’t official but every credible source is pointing to massive revenue growth
define “credible”
Also do you think they would cook the books to that extent if they are still going ahead with an IPO? Fabricating $50b in revenue would hurt an IPO once it’s found out way more then some temporary hype from a NYT article.
Well unless they have reach a level where training is not needed (which will be never for an LLM), what you told me is that you make a profit freezing ice cream but lose it on the eggs, cream and sugar necessary to make it
The model releases are getting shorter and shorter. close to a month now. There is no actual profitability not because the price isn’t higher than cost of ingredients, but because they don’t count throwing away the ice cream mixer, and factory every month, when they should if replacing them is part of the process.
Training is not needed to run inference on the existing models, it’s needed to make new models. If LLMs plateau and there’s no use in training new models, or the government regulates them and they can’t train new frontier models, then they can still run inference on the currently existing models and make a profit.
This is more like making a profit for making ice cream but losing a bunch of money researching new recipes for ice cream. If you decide the recipe is good enough and stop doing that research you can still make money making and selling ice cream.
yeah? so a model trained in 2 year old data is just as good? as a model trained today?
So they are just choosing to lose money? they have a perfectly good, profitable product but they choose to lose money… ok bud
Your point only makes sense if we agree the current models are still not good enough to make a profitable product and thus, they need to keep pushing
Yes, assuming a plateau in LLM capability the only reason you’d train on an updated corpus is to get fresh info. That’s not worth it because:
Most of the improvement from new models isn’t coming from expanding or updating the corpus, its coming from increasing the parameter size and reinforcement learning.
No, they are competing. If open AI decides to stop training new models right now then anthropic will and take all there business as the switching cost for models is low. That’s why they want the government to regulate it, so they can have a ceasefire to start taking in profits from inference without worrying about their competitor making a new better model and eating there lunch.
if that were true, and the current models were actually good, there would be wide adoption of the current models while people wait for the next leap… instead, you get glimpses of the terrible numbers like 3% of user base is actually paying for these things
maybe we torture our analogies to death but you make it sound like they have the formula for coca cola but instead of selling that, they are burning cash trying to find an even better formula
when the reality is that they have something that is technically drinkable and not officially poison and they are selling it like the next elixir of life (while they burn cash trying to catch up to their own hype)
People are adopting the current models, mostly coders right now. Most people using LLMs are using them for basic information search, ie. Google AI overviews. Those tasks don’t require the top models and don’t burn that many tokens so the companies keep them free to get the public exposed to AI.
Then there’s the 3% of people who are the power users using it for work, especially coders. For productivity the top models do perform better and the stakes are higher so they need to perform better. The top models aren’t needed for every task, but they shine as an orchestrator of smaller models handling more basic tasks. Coding also requires a lot more tokens, to read all the existing code; to do “thinking” which generates a bunch of output tokens to imitate reasoning; then to generate the code itself; then to review it, adjust after a review…
All of that equates to large bills for token spend to anthropic or open AI. My Claude bill on my company account is approaching $2,000 this month, and that’s about average / what’s expected from an engineer at my company. I’d bet that every engineer in silicon valley is burning through a similar amount as well.
This is why both open AI and anthropic are both seeing extremely high revenue growth. Anthropic ended 2025 with $9b in revenue , they are now on track to hit $100b revenue this year, for reference / the analogy coca cola made $47b this year, it’s a larger revenue then every other software company except Microsoft. THAT IS INSANE, no company has ever seen that kind of growth. Yes it’s being weighed down by training costs so they aren’t profitable but if they even double there revenue next year that could change.
Models are being push down people’s throats, mostly to coders right now… there FIFY
Again, this is forced, you literally cannot perform a google search (or bing) without getting some AI results whether you want it or not. Then AI companies, like you, use this to pretend it’s actual demand for their products
If 3% of people generate the cost the AI companies cannot keep to the prices they charge, then this is the definition of an unprofitable business
No, this is what they ARE CLAIMING THEY WILL REACH. Last year, Altman said it was ridiculous that people were saying they generated $12 billion in revenue… just to then show it was less than that. Now they are delaying their IPO which means the books stay closed for another while longer
Again you are just drinking the koolaid here (not counting all the anecdotal evidence you now apply to all of Silicon Valley). The truth is that whatever revenue they actually have (or claim, “trust me bro”), you can’t even know if half of it comes from new investments, creative accounting or selling off assets, you just assume it is more people demanding more of their services and paying through the nose for it even if they have barely showed any tangible results in real life https://fortune.com/2026/08/22/executives-ai-productivity-layoffs-study/
Oh that’s it? just double the already inflated revenue they are reporting without any evidence? of course!!! and all I need to fly is to grow some wings, that’s it, it’s really that simple
I don’t think you understand the scale, even if half of that number is real that’s $50b , that’s larger then coca cola, that’s larger then Salesforce and every other software company except Microsoft. That’s 5x growth on $10b , that is unheard of in the software industry, much less any other industry.
IDK what to tell you if you don’t believe the numbers, yes they aren’t official but every credible source is pointing to massive revenue growth. We’ll see the actual numbers in the coming weeks as they prepare for the IPO, and if they do turn out to be cooking the books to the tune of $50b , I’ll come back to this comment and admit I’m a credulous idiot who fell for the hype. But if they are true I’ll also be coming back with an I told you so.
Also do you think they would cook the books to that extent if they are still going ahead with an IPO? Fabricating $50b in revenue would hurt an IPO once it’s found out way more then some temporary hype from a NYT article.
and you only need the to double it (the full 100 billions} to make a profit lol…
define “credible”
the ipo they are already postponing for “safety”?
https://fortune.com/2026/09/12/sam-altman-openai-ipo-delay-ill-advised-moment-safety-concerns/
https://www.investors.com/news/technology/anthropic-ipo-delayed-openai-expects-massive-cash-burn/