CEOs of companies that are adjacent to technology desperately want to ensure that their company isn't seen as "outdated", almost more than they want to actually not be outdated.
So when a technology comes that everyone in tech leadership is saying is the bestest, they want to make sure everyone knows they're totally with it, whatever the cool kids are talking about.
Hype train goes chugga chugga.
As the hype train slows, they still need to be onboard, but they set expectations based on what their people are actually telling them.
So this is the CEO yelling to do something, and then the news slowly percolating back from the tech people that they can, but only a handful of projects can do so in a way that makes sense, has impact, and doesn't disrupt a timeline or budget in a way that requires shareholder disclosure.
A lot of tech firms bought into the gold rush mindset. NVIDIA sold the shovels. Sooner or later, when they figure out the world isn't excitedly opening their wallets, the bubble will pop, and they'll move on to the next empty hype train.
This is it. The real product is hype, with a tiny tiny little kernel of actual utility, that is puffed up and remixed until the hype dies away, and we have to make do with whatever's left.*
The hype machine with generative shit went into fuckin overdrive because while yes there is a grift component, natch; unlike with blockchain, nfts, web3, etc, there is an actual visible thing that the technology can do that hasn't been done before.
People who are used to selling nothing but vapour lost their minds when they saw it, because rightfully so, they realised how much grift they were going to be able to make off of it.
* Usually this involves a bunch of platform engineers et al de-tooling codebases and infrastructure.