There’s a financial bubble and a mental bubble. They are related and feed each other but they are not the same and might not blow up at the same time.
The Market, according to the article, isn’t losing its faith in AI as the Ultimate Technology (the mental bubble or “AI psychosis”). It’s just questioning its belief that one or two companies might be able to squeeze massive profits out of it (the financial bubble).
Investors thought one AI winner would emerge relatively quickly, a new SuperGoogle or HyperAmazon that would make all the money in the world by providing access to AI to every other company. The AI winner would then gleefully enshittify their product while making it more expensive over time, eventually turning huge profits (even including the Trillions of investment spent to get there).
Now, with multiple companies that are offering an almost equivalent product and are forced to compete on price while already bleeding billions every quarter, everyone might just run out of money before the winner is declared. Not a scenario many are willing to invest in.
But the idea that AI can do everyone’s job or give anyone 100x superpowers will still be out there. And all the CEOs and CTOs who thought they would look like thought leaders if they bet everything on AI will not easily change their minds all at once. They will still be easy marks for the next snake oil salesman, selling them a product based on cheap, less powerful models or whatever they can insist can still perform the 100x miracle at an actually bearable cost.
Some might actually be useful products, it’s just that if the mental bubble, overinflating what AI can do doesn’t blow, they can still be sold way above their actual worth.
Yes, but them spending a lot of company money for a low ROI product while sacking experienced employees will have a detrimental effect on their company’s performance as well. And thus on their own stock value.
If you are an investor, this is a classic picks and shovels play. No one is going to be able to guess which lab puts out the best models, none of them really have a moat. However, every lab is going to need massive amounts of hardware to train and serve their models. It doenst even matter if open source models end up winning, they still need massive hardware to train and infer.
There’s a financial bubble and a mental bubble. They are related and feed each other but they are not the same and might not blow up at the same time.
The Market, according to the article, isn’t losing its faith in AI as the Ultimate Technology (the mental bubble or “AI psychosis”). It’s just questioning its belief that one or two companies might be able to squeeze massive profits out of it (the financial bubble).
Investors thought one AI winner would emerge relatively quickly, a new SuperGoogle or HyperAmazon that would make all the money in the world by providing access to AI to every other company. The AI winner would then gleefully enshittify their product while making it more expensive over time, eventually turning huge profits (even including the Trillions of investment spent to get there).
Now, with multiple companies that are offering an almost equivalent product and are forced to compete on price while already bleeding billions every quarter, everyone might just run out of money before the winner is declared. Not a scenario many are willing to invest in.
But the idea that AI can do everyone’s job or give anyone 100x superpowers will still be out there. And all the CEOs and CTOs who thought they would look like thought leaders if they bet everything on AI will not easily change their minds all at once. They will still be easy marks for the next snake oil salesman, selling them a product based on cheap, less powerful models or whatever they can insist can still perform the 100x miracle at an actually bearable cost.
Some might actually be useful products, it’s just that if the mental bubble, overinflating what AI can do doesn’t blow, they can still be sold way above their actual worth.
It will peak and become more boring and Google like.
Yes, but them spending a lot of company money for a low ROI product while sacking experienced employees will have a detrimental effect on their company’s performance as well. And thus on their own stock value.
If you are an investor, this is a classic picks and shovels play. No one is going to be able to guess which lab puts out the best models, none of them really have a moat. However, every lab is going to need massive amounts of hardware to train and serve their models. It doenst even matter if open source models end up winning, they still need massive hardware to train and infer.