Ed Zitron (search up his blog) has some excellent and extensive writing on why this is happening. Cory Doctorow also covers a lot of it.
The short version is this; corporations no longer exist to be successful by any normal metrics (ie, you have lots of customers that you sell lots of product to). They exist to serve shareholders. Shareholders don’t want stability, they want growth. They want a stock that will 2x or 3x so they can dump it and make a killing. So everything is about growth. No one cares how big you are today, they only care how big you will be tomorrow.
This is especially bad in tech, because tech was the growth sector for the past forty years. But there’s no growth left. Almost everyone who can be online is online. Microsoft, Amazon, Google and Facebook have signed up every customer they possibly can. There are no opportunities for explosive new growth left. This is what has lead to enshittification; they can’t bring in new customers, so now they have to squeeze their existing customers by upping prices, cutting services, and cramming ads into everything.
Then OpenAI showed off ChatGPT and the tech companies lost their shit. Here, finally, was something new, something big. Something that promised meteoric growth. Something with limitless untapped potential. Entire industries erased, entire workforces made redundant, all replaced with a product that Microsoft, Amazon, Google or Facebook/Meta can sell to millions of companies. The sky was the limit.
So they all shovelled billions of dollars into AI, either their own projects or startups that are functionally wholly owned subsidiaries. And they told their shareholders that these billions were worth it for the untold riches that would rain down on them all very soon.
And then it turned out that ChatGPT was not the seed of something, but rather the withered old husk. LLMs are, functionally, a dead end. They cost so much power to run that there’s literally no way to make them profitable, and every attempt to solve their deep shortcomings involves strapping more of them together in giant AI voltrons, upping the already ridiculous operating costs.
For example OpenAI currently spends $2.35 to make $1.00. That’s not one dollar of profit. That’s one dollar. Period. Yes, that means they lose a dollar thirty five in that exchange. And that’s with the ridiculously subsidized compute rates they get from Microsoft. The flagship gen AI name is a horrendously unprofitable mess.
AI might still have vast and explosive growth potential, but the current methods will not get us there, a fact that is becoming increasingly clear to everyone.
Which leaves the big tech companies with a serious problem. They promised untold riches, and if they do not deliver their shareholders are going to absolutely fucking eviscerate them. It will be a bloodbath. So they have to either come clean now and get dragged through the streets, or they have to double down, keep promising that the explosive growth is right around the corner, and keep spending money in order to signal just how incredibly confident they are that this bet is going to pay off. They’re going to take every AI product they can cobble together and force it onto their customers so they can produce fabricated metrics for their shareholders demonstrating massive and steadily growing demand. And all the while they’re going to pray that maybe, just maybe, these overconfident engineers they’ve been handing dump trucks full of money to will pull a rabbit out of a hat and actually deliver the promised miracle tech.
Ed Zitron (search up his blog) has some excellent and extensive writing on why this is happening. Cory Doctorow also covers a lot of it.
The short version is this; corporations no longer exist to be successful by any normal metrics (ie, you have lots of customers that you sell lots of product to). They exist to serve shareholders. Shareholders don’t want stability, they want growth. They want a stock that will 2x or 3x so they can dump it and make a killing. So everything is about growth. No one cares how big you are today, they only care how big you will be tomorrow.
This is especially bad in tech, because tech was the growth sector for the past forty years. But there’s no growth left. Almost everyone who can be online is online. Microsoft, Amazon, Google and Facebook have signed up every customer they possibly can. There are no opportunities for explosive new growth left. This is what has lead to enshittification; they can’t bring in new customers, so now they have to squeeze their existing customers by upping prices, cutting services, and cramming ads into everything.
Then OpenAI showed off ChatGPT and the tech companies lost their shit. Here, finally, was something new, something big. Something that promised meteoric growth. Something with limitless untapped potential. Entire industries erased, entire workforces made redundant, all replaced with a product that Microsoft, Amazon, Google or Facebook/Meta can sell to millions of companies. The sky was the limit.
So they all shovelled billions of dollars into AI, either their own projects or startups that are functionally wholly owned subsidiaries. And they told their shareholders that these billions were worth it for the untold riches that would rain down on them all very soon.
And then it turned out that ChatGPT was not the seed of something, but rather the withered old husk. LLMs are, functionally, a dead end. They cost so much power to run that there’s literally no way to make them profitable, and every attempt to solve their deep shortcomings involves strapping more of them together in giant AI voltrons, upping the already ridiculous operating costs.
For example OpenAI currently spends $2.35 to make $1.00. That’s not one dollar of profit. That’s one dollar. Period. Yes, that means they lose a dollar thirty five in that exchange. And that’s with the ridiculously subsidized compute rates they get from Microsoft. The flagship gen AI name is a horrendously unprofitable mess.
AI might still have vast and explosive growth potential, but the current methods will not get us there, a fact that is becoming increasingly clear to everyone.
Which leaves the big tech companies with a serious problem. They promised untold riches, and if they do not deliver their shareholders are going to absolutely fucking eviscerate them. It will be a bloodbath. So they have to either come clean now and get dragged through the streets, or they have to double down, keep promising that the explosive growth is right around the corner, and keep spending money in order to signal just how incredibly confident they are that this bet is going to pay off. They’re going to take every AI product they can cobble together and force it onto their customers so they can produce fabricated metrics for their shareholders demonstrating massive and steadily growing demand. And all the while they’re going to pray that maybe, just maybe, these overconfident engineers they’ve been handing dump trucks full of money to will pull a rabbit out of a hat and actually deliver the promised miracle tech.