The AI boom has turned the standard profit margin model on its head, according to Apollo Chief Economist Torsten Slok—and it’s making the industry’s growth unsustainable.
The underlying technology isn’t unprofitable, inference alone isn’t unprofitable. It’s the research and development of better models that’s unprofitable and unsustainable.
Investment IS gambling, some people will win, some will lose.
That’s pretty much how almost every new development happens. EVs were the same, primarily funded by subsidies and now they’re cost competitive or better in many situations even without them. Cellphone networks… also heavily funded by investors and some took hits from how it all played out. The dotcom bubble. Even as far back as the Railroad boom in the UK. This is a tale as old as time.
The underlying technology isn’t unprofitable, inference alone isn’t unprofitable. It’s the research and development of better models that’s unprofitable and unsustainable.
Investment IS gambling, some people will win, some will lose.
That’s pretty much how almost every new development happens. EVs were the same, primarily funded by subsidies and now they’re cost competitive or better in many situations even without them. Cellphone networks… also heavily funded by investors and some took hits from how it all played out. The dotcom bubble. Even as far back as the Railroad boom in the UK. This is a tale as old as time.
Yes it is. The only things its useful and efficient for are all laid out in either ‘1984’ or the Geneva conventions.
Your emotions don’t override reality.
LLMs are excellent at translation as one significant beneficial use case.
I wouldn’t put a fancy and stupidly resource-intensive word guessing game on the same level as those other things.
Maybe you wouldn’t, but I would. You clearly don’t remember how bad the internet was when it first became available to the public.
Do tell, how has it improved since the times that I clearly don’t remember?