Jun 2, 2026 · 26 min · 14 segments
**The enterprise AI bubble just hit its sharpest fiscal reality check—and Uber is leading the panic.** In Episode 80 of *What the AI?!*, Jeff Keltner and Annie Delgado break down a massive economic…
Annie DelgadoHostUm, and the COO was kind of out there in an interview going, um, "Is this, is this really-- Are we delivering anything? It's not clear that we're getting the benefit we expect from all the money we're spending on these AI coding tools." That was a really interesting, I thought, um, concept and, and idea that he was putting out there.
I mean, normally you hear a lot larger numbers when you're talking about AI, and so four is, four is quite a small number.
But it, it is kind of staggering to think, like, companies exited 2025 expecting to spend millions of dollars in 2026 on AI tools, and actually-
they spent all of those millions in the first quarter or the first, you know, four months, whatev-whatever it is.
I think it, it's more of the tension that we talked about last week, where it's like you're spending a lot of money on essentially training, 'cause people don't know how to turn the tools into actual productivity yet until they actually learn how to use them.
It is a large training budget, and it's, you know, uh, even those of us who use the tools a lot, like there are lots of instances where it kind of sort of helps, but the amount of time you spend overseeing and, "Uh, not quite right.
Do it like this," and mm, and then double-checking, and you go, you know, it might've been, you know, might've been easier time-wise and less expensive to have at the hourly rate of the person doing the double-checking versus the dollars per token and the number of tokens involved in these agentic models.
the number of dollars per intelligent insight that being generated or code line is not trivial.
And so it's just not clear, at least right now, that the economics work out in a lot of these cases to pay for the AI to do the work instead of just having the, the humans do it [chuckles].
I'm, I'm really curious at one, at what point we start to see the CFOs step in and say, "All right, we're a little off the rails here, guys.
Um, and the COO was kind of out there in an interview going, um, "Is this, is this really-- Are we delivering anything? It's not clear that we're getting the benefit we expect from all the money we're spending on these AI coding tools." That was a really interesting, I thought, um, concept and, and idea that he was putting out there.
I mean, normally you hear a lot larger numbers when you're talking about AI, and so four is, four is quite a small number.
But it, it is kind of staggering to think, like, companies exited 2025 expecting to spend millions of dollars in 2026 on AI tools, and actually-
they spent all of those millions in the first quarter or the first, you know, four months, whatev-whatever it is.
I think it, it's more of the tension that we talked about last week, where it's like you're spending a lot of money on essentially training, 'cause people don't know how to turn the tools into actual productivity yet until they actually learn how to use them.
It is a large training budget, and it's, you know, uh, even those of us who use the tools a lot, like there are lots of instances where it kind of sort of helps, but the amount of time you spend overseeing and, "Uh, not quite right.
Do it like this," and mm, and then double-checking, and you go, you know, it might've been, you know, might've been easier time-wise and less expensive to have at the hourly rate of the person doing the double-checking versus the dollars per token and the number of tokens involved in these agentic models.
the number of dollars per intelligent insight that being generated or code line is not trivial.
And so it's just not clear, at least right now, that the economics work out in a lot of these cases to pay for the AI to do the work instead of just having the, the humans do it [chuckles].
I'm, I'm really curious at one, at what point we start to see the CFOs step in and say, "All right, we're a little off the rails here, guys.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.