Oct 21, 2026 · 53 min · 13 segments
AI adoption is replaying many of cloud’s early mistakes, but agentic systems introduce a broader spending population, nondeterministic costs, and systems that can act on their own. In this sponsored…
Eyvonne SharpHostAnd the first question I'd like to go with is, I've personally spent a ton of time, especially recently, thinking about the early days of enterprise cloud adoption.
And, you know, I worked for some big companies that, you know, decided that they were going to go all in, you know, kind of like reduce that data center footprint and go all in the cloud and maybe even decom the data centers, you know, at some point if they could.
And, you know, that experience years and years ago um taught us some maybe some of us some expensive lessons about costs about governance and uh ultimately about operating models like how do we get to the right operating model and i'm kind of seeing a very similar it's deja vu almost like a similar pattern with ai where orgs are moving first and kind of trying to figure out how to operate what they've built afterwards, maybe.
And it doesn't help that this stuff is changing so fast, like almost on a day-to-day basis, it seems like.
But I guess a question for both of you or one of you or whoever wants to jump in, where is the comparison between early cloud adoption and today's AI adoption accurate? Where is it really accurate? And where does AI create I guess, fundamentally different problems, if that makes sense.
Yeah, I'll start because I have a feeling Ange probably got even more when it gets into the financial ends of things.
But yeah, there is a lot of that sort of, you know, we had those shadow IT kind of considerations coming up just 20 years ago this last month, I think, or August 2020, 2006 was when some of the first cloud capabilities came out.
So there was always this challenge inside of organizations of trying to strike that balance between the innovation and wanting to kind of go and explore new things but also trying to protect their organizations as well those governance kinds of aspects so what happened in the cloud era was a lot of the sort of like you know adopt first and we bolt on the governance and the fin ops and everything else at a later point in time and We could say that that really maybe wasn't the best approach, but it's kind of where we landed on things.
AI is following similar patterns the way I see it, but I actually think that the blast radius on that is a little bit larger, potentially, because whereas we might have something out in a cloud, we don't like it out there or something like that, when you start to get into agentic types of things, we're going to have AIs that are actually... executing things and making decisions for our business.
We've seen some examples of this already in some areas where it's like, okay, well, the AI made a decision for the business and now the business has to deal with that.
And then from my side, just thinking about some of the financial aspects and how FinOps has matured over the years, we're definitely in the AI FinOps and cost world, kind of where we were about maybe 10 years ago in terms of cloud adoption.
So a lot of people struggling with just basic visibility now, trying to figure out who's spending what.
But one of the things that for me to Doug's points about blast radius that I definitely see on the AI FinOps side is the core of cloud FinOps was really about being able to spread expertise, spread accountability and responsibility.
And the difference between AI and cloud and that particular element is that When I'm doing cloud, I'm spreading this to people who are requesting sizes of servers, requesting specific services and things like that.
They have an amount of technical expertise that at the very least, they know what a server is.
But as I get to AI and I sort of get to AI FinOps, I'm now spreading that accountability that I have to build out to people that Maybe they know what Microsoft Office is.
They don't necessarily know that a server costs a whole ton of money or the memflation crisis, things like that.
That's all outside their wheelhouse, but the choices that they make with the way that they use AI have material effects in terms of cost.
So the technical capacity of the user of AI is far less than the average technical capacity of somebody who uses cloud.

And I'm thinking here, you know, an example would be, you know, you may have AI power users in your organization who can't really even tell you what a token is, right? Like that's kind of the arena we're talking about.

So how does this change how organizations think about tokens? cost management, accountability, and how to even design a practice around it, because I imagine some of those cloud models aren't going to map directly.
And the first question I'd like to go with is, I've personally spent a ton of time, especially recently, thinking about the early days of enterprise cloud adoption.
And, you know, I worked for some big companies that, you know, decided that they were going to go all in, you know, kind of like reduce that data center footprint and go all in the cloud and maybe even decom the data centers, you know, at some point if they could.
And, you know, that experience years and years ago um taught us some maybe some of us some expensive lessons about costs about governance and uh ultimately about operating models like how do we get to the right operating model and i'm kind of seeing a very similar it's deja vu almost like a similar pattern with ai where orgs are moving first and kind of trying to figure out how to operate what they've built afterwards, maybe.
And it doesn't help that this stuff is changing so fast, like almost on a day-to-day basis, it seems like.
But I guess a question for both of you or one of you or whoever wants to jump in, where is the comparison between early cloud adoption and today's AI adoption accurate? Where is it really accurate? And where does AI create I guess, fundamentally different problems, if that makes sense.
Yeah, I'll start because I have a feeling Ange probably got even more when it gets into the financial ends of things.
But yeah, there is a lot of that sort of, you know, we had those shadow IT kind of considerations coming up just 20 years ago this last month, I think, or August 2020, 2006 was when some of the first cloud capabilities came out.
So there was always this challenge inside of organizations of trying to strike that balance between the innovation and wanting to kind of go and explore new things but also trying to protect their organizations as well those governance kinds of aspects so what happened in the cloud era was a lot of the sort of like you know adopt first and we bolt on the governance and the fin ops and everything else at a later point in time and We could say that that really maybe wasn't the best approach, but it's kind of where we landed on things.
AI is following similar patterns the way I see it, but I actually think that the blast radius on that is a little bit larger, potentially, because whereas we might have something out in a cloud, we don't like it out there or something like that, when you start to get into agentic types of things, we're going to have AIs that are actually... executing things and making decisions for our business.
We've seen some examples of this already in some areas where it's like, okay, well, the AI made a decision for the business and now the business has to deal with that.
And then from my side, just thinking about some of the financial aspects and how FinOps has matured over the years, we're definitely in the AI FinOps and cost world, kind of where we were about maybe 10 years ago in terms of cloud adoption.
So a lot of people struggling with just basic visibility now, trying to figure out who's spending what.
But one of the things that for me to Doug's points about blast radius that I definitely see on the AI FinOps side is the core of cloud FinOps was really about being able to spread expertise, spread accountability and responsibility.
And the difference between AI and cloud and that particular element is that When I'm doing cloud, I'm spreading this to people who are requesting sizes of servers, requesting specific services and things like that.
They have an amount of technical expertise that at the very least, they know what a server is.
But as I get to AI and I sort of get to AI FinOps, I'm now spreading that accountability that I have to build out to people that Maybe they know what Microsoft Office is.
They don't necessarily know that a server costs a whole ton of money or the memflation crisis, things like that.
That's all outside their wheelhouse, but the choices that they make with the way that they use AI have material effects in terms of cost.
So the technical capacity of the user of AI is far less than the average technical capacity of somebody who uses cloud.

And I'm thinking here, you know, an example would be, you know, you may have AI power users in your organization who can't really even tell you what a token is, right? Like that's kind of the arena we're talking about.

So how does this change how organizations think about tokens? cost management, accountability, and how to even design a practice around it, because I imagine some of those cloud models aren't going to map directly.
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.