Here’s the problem…
He says AI was adopted beyond his expectations. Great.
But if somebody is using it at the current price point of super cheap or free, are they going to keep using it when it gets expensive?
You can make a basic chatbot run on a desktop PC, but nobody wants to pay for that. Once you get into things with useful generation and large context windows, or things like video generation, suddenly you need one or more $10,000+ pieces of hardware to run it.
So the $10 a month you charge the user is basically an introductory price that doesn’t cover your hardware fees let alone the software engineers to build your AI.
Eventually, the bill comes due. Eventually, you have to look at your customers and how much machine time they use each month and how much your r&d costs and figure out what the actual cost to the customer has to be. And then the customer rethinks how useful the AI is or isn’t.
People will pay $10 a month for chat GPT to write their emails. Will they pay $100 a month?
What about the company that replaced all their software developers with AI. Suddenly the AI cost as much or more as the software developers. Only now the developers who understood the code base work for other companies.
People will pay $10 a month for chat GPT to write their emails.
They won’t even pay that in many cases but instead rely on the free offerings. Expect those to go away altogether, to be heavily hamstring, or to become absolutely riddled with advertisements.
Once you get into things with useful generation and large context windows, or things like video generation, suddenly you need one or more $10,000+ pieces of hardware to run it.
A Blackwell server with 72 GPUs costs about $3 million, plus requires 130 kW of power (about 3 residential homes’ max rated power through a residential 200A circuit box, for about $600-$1000/day in electricity cost).
You’re gonna need to sell a lot of $20/month subscriptions to get that paid for, assuming that the server is good for 5 years. If it’s only good for 3 years, the economics are basically impossible.
… assuming that the server is good for 5 years. If it’s only good for 3 years, the economics are basically impossible.
And surely the big AI companies wouldn’t decide that hardware rated for ~3 years should be amortized over 5-6 years in order to massively inflate their value on paper.
I’m sure Bernie Madoff’s fund had an ‘accounting puzzle’ too.
I love how the publications call it a ‘puzzle’ rather than what it so obviously is- a bunch of people throwing $billions at tech they don’t understand which has no serious road to profitability anytime soon.
If it’s only good for 3 years, the economics are basically impossible.
Also consider that as models improve, the newer frontier models that are doing actual useful work require significantly more compute and RAM than basic chat bots…
Here’s the problem… He says AI was adopted beyond his expectations. Great.
But if somebody is using it at the current price point of super cheap or free, are they going to keep using it when it gets expensive?
You can make a basic chatbot run on a desktop PC, but nobody wants to pay for that. Once you get into things with useful generation and large context windows, or things like video generation, suddenly you need one or more $10,000+ pieces of hardware to run it. So the $10 a month you charge the user is basically an introductory price that doesn’t cover your hardware fees let alone the software engineers to build your AI.
Eventually, the bill comes due. Eventually, you have to look at your customers and how much machine time they use each month and how much your r&d costs and figure out what the actual cost to the customer has to be. And then the customer rethinks how useful the AI is or isn’t.
People will pay $10 a month for chat GPT to write their emails. Will they pay $100 a month?
What about the company that replaced all their software developers with AI. Suddenly the AI cost as much or more as the software developers. Only now the developers who understood the code base work for other companies.
There will be a fun correction when this happens.
They won’t even pay that in many cases but instead rely on the free offerings. Expect those to go away altogether, to be heavily hamstring, or to become absolutely riddled with advertisements.
A Blackwell server with 72 GPUs costs about $3 million, plus requires 130 kW of power (about 3 residential homes’ max rated power through a residential 200A circuit box, for about $600-$1000/day in electricity cost).
You’re gonna need to sell a lot of $20/month subscriptions to get that paid for, assuming that the server is good for 5 years. If it’s only good for 3 years, the economics are basically impossible.
And surely the big AI companies wouldn’t decide that hardware rated for ~3 years should be amortized over 5-6 years in order to massively inflate their value on paper.
Surely.
I’m sure Bernie Madoff’s fund had an ‘accounting puzzle’ too.
I love how the publications call it a ‘puzzle’ rather than what it so obviously is- a bunch of people throwing $billions at tech they don’t understand which has no serious road to profitability anytime soon.
Also consider that as models improve, the newer frontier models that are doing actual useful work require significantly more compute and RAM than basic chat bots…