FinOps X 2026 | Robbie Plourde, USU

Clips
More from FinOps X 2026

Robbie Plourde

Principal Solution Engineer

USU

play_circle_outline Opaque AI Tokenomics: How Hidden Token Costs Distort FinOps Unit Economics

play_circle_outline Hybrid and On‑Prem Comeback: Data Locality, Foundational Data Quality, and Translating Costs into Business Value

play_circle_outline Governing AI Costs: Sandboxed Innovation, Production Controls, and EU Data Sovereignty

play_circle_outline FinOps community growth and convergence around tokenomics and cost practices

Info
Transcript

Robbie Plourde, USU

Robbie Plourde

Principal Solution Engineer USU

In this interview from FinOps X 2026, Robbie Plourde, executive director of sales and business development at USU Solutions, joins theCUBE's John Furrier and theCUBE Research's Paul Nashawaty to discuss the evolution of FinOps from cloud cost control into an enterprise discipline for AI value management and token economics. Drawing on a traditional IT asset management background, Plourde explains how token-based pricing obscures costs in ways that compute and storage never did — a single opaque line item that engineers consume without tracking the downstream ... Read more

explore Keep Exploring

How is FinOps evolving with the emergence of AI and tokenomics, particularly regarding unit economics, cost transparency, and tying costs to business value? add

How does the shift to hybrid cloud and on‑premises AI/SaaS deployments change the data and cost-management equation and the role of practitioners? add

How is Europe approaching AI regulation compared with the United States, and what are the implications for compliance, cross-border applications, innovation (sandboxing), penalties, and costs? add

What are you optimizing for this year, and what is your focus? add

bolt Powered by CUBE AI

Robbie Plourde, USU

search

John Furrier

>> Welcome back to theCUBE's livestream here in San Diego for FinOps X 2026. I'm John Furrier, host of theCUBE with my coast, Paul Nashawaty with theCUBE Research. Robbie Plourde here, principal solutions engineer of USU. A lot of cloud management, a lot of cost management with AIs here. Robbie, thanks for joining us.

Robbie Plourde

>> Thank you for having me.

John Furrier

>> Appreciate coming on. First of all, the AI value is a top story. Okay, obviously tokenomics, big part of the theme because of the foundation announcement. FinOps is changing. How do you see that evolution? Because it's very natural. Got great reviews, by the way. It made sense. First, you're like, wait a minute, a new foundation? It's all kind of coming together. What's your view?

Robbie Plourde

>> It is. Yeah. Coming from a traditional ITAM background, we see some similarities on when we went from on-premise to cloud, the shift that was made there. What we see is that when you look at the traditional way of doing the unit economics within cloud, you have particular items like compute storage, which were pretty easy and understood and very clear in your billing cycle. When you start looking at AI and then bring in tokens, you have the price on your sheet, but you really don't know how to break down that price and how to differentiate how that price should be set against different things. So basically, you have the demand there, but not how to justify that demand back to the business or where it ties back into the business. So that's where we really have to figure out how to get that hidden cost and that opaque cost front and center. So that's the difference is it's still foundational to understand the data, but it's the business value that you apply to that is really where you have to start doing that.

John Furrier

>> And the AI spend obviously is on everyone's mind, but now you have cross environment in a distributed manner, aka hybrid cloud.

Robbie Plourde

>> Yeah, exactly.

John Furrier

>> Now the on premises data centers are starting to emerge. Why? Because that's where the data is.

Robbie Plourde

>> Exactly.

John Furrier

>> And so there's still migrations to the cloud for obvious reasons, but on prem is active. It is. And you got SaaS on prem, you got AI on prem and everyone's adopting AI. How does that change the equation?

Robbie Plourde

>> It changes it just in the manner of the data that you're looking at and the data that you're collecting. Data's always been the foundation. And when you bring AI in, your AI is only as good as your foundation data, right? AI is going to give you a confidently wrong answer if you don't have very good data to back that up, especially around cost. When you look at the cost data, if it's a mess, your AI results, again, are going to be confidently wrong and you have to basically understand how to interpret that data and that's where the role of the practitioner has changed. When you go from collecting the data, you're now looking at, you don't have to really collect it, you have to understand it, interpret it into a business value, and that is where things are kind of shifting right now.

Paul Nashawaty

>> Let's talk about that shift. I mean, there's a ways to look at the data. There's ways to look at the information, the results and such. In our research, we see that organizations are using six to 15 different tools in order to understand what's going on here. That's an amazing amount of tools. It's amazing amount of overhead, right?

Robbie Plourde

>> Yeah.

Paul Nashawaty

>> And 54% of those respondents indicated they're looking in favor of moving towards a unified view. So with that said, we have these silos, we have these individual ways to looking and diagnose the issues of the cost of the usage and such. What are your thoughts around the tools and the tool sprawl and then how do people normalize that and harmonize that?

Robbie Plourde

>> Yeah. The tool sprawl has come because different departments need the data to do different things and they're all using different tools that feed their needs. What we need to see is a convergence of not so much the technology stack but the understanding of the data behind it to say, "We're all using the same data. Let's find a tool that maybe collects the data and gives us that information in different ways that we can then bring back to the business and give that business value to that because the shift has been away from cost reduction to cost validation or understanding your cost rather than trying to reduce it." So when you have all of these different technologies and different tools collecting the data, to normalize them all under one tool, it's going to be difficult, right? Because finance needs something different than the IT operations team. And then technologists need the data in a different format. So the tool sprawl is really... I don't know how you solve that.

Paul Nashawaty

>> It is. It is. It's a problem, but there's a lot of vendors coming out with AI-optimized tools for fin ops, right?

Robbie Plourde

>> Yes.

Paul Nashawaty

>> There's tools that allow for this fix, right?

Robbie Plourde

>> Yes.

Paul Nashawaty

>> But there's a challenge between offering up these features and doing it yourself.

Robbie Plourde

>> Correct. Yeah.

Paul Nashawaty

>> You build it yourself because, like you said, you have specific needs. I mean, obviously there's a faster time to value. There's you don't have to own it, that kind of thing. But besides the operational costs around it, what's the advantage of going with an AI feature set that a vendor is providing versus a DIY?

Robbie Plourde

>> Yeah. If you're getting one that's embedded into the tool, you're getting the value of the building of the model, if you will, right? The understanding of cross data of different customers getting the data and getting that. Building it your own, you're only using your data. And do you have the engineering team the time to build up that model and to train it? Building in the tool that I... The company I work for has a tool that has AI embedded in it, right? And we learned it's not as easy as it looks to... One, building it's easy. When you train that, we dumped in millions of invoices into this tool to train it. And all of the time behind that, the engineering team's like, "I don't have time to train the tool now. I built it." So when you have a tool embedded or AI functions embedded into a tool, you get that value from the vendor rather than building it yourself, but you lose out on something such as a closed model and then you don't understand if you're getting it from the vendor, you don't know how it's working, and that's the benefit against the downside of it.

John Furrier

>> Well, when someone asks the question, "How much return am I getting per dollar from the AI?" what is the current approach?

Robbie Plourde

>> No one is really doing that very well at this point because they have data and they can do it per something, the unit economic if you will. And what the shift in token is doing is it hides that cost. So now, in order for you to say, "How much is the cost per something?" it's a one line item that's very difficult for you to break up and understand because the engineering team is creating the demand without the understanding of how that's being calculated in the backend on demand. It's based on the model. It's based on something else. So it's very hard to control that cost and understand where that's coming from.

Paul Nashawaty

>> Is that intentional? I mean, think of what it went-

Robbie Plourde

>> I feel like it is.

Paul Nashawaty

>> Yeah, because when cloud was first coming out, it was a, "Here's your line item." And then it was like, no, no, we got to break that apart. Now we have this token that's like, "Here's your line item. I feel like it needs to be broken apart."

Robbie Plourde

>> It is. It's a way to hide the cost, if you will. It's very easy. It's almost like if you go to an amusement park now, they give you the card and they say, "Just use this. Don't take out your credit card. Just use this." You never know how much you're really spending until you get to the end. And this is the same idea is, "Hey, we'll let them consume as many tokens as they want." And we learned early on they were getting large amounts of tokens, not understanding that you're going to burn through those in a month instead of a year, which you've quoted because, like I said, the engineers are the ones using those tokens and they're not understanding, "Hey, if I choose this model, it's going to be more than the other one." So they hide the costs, they get you hooked on it until you really have to understand that. That's that build versus buy approach as well. You have control over that, but the vendors don't want you to understand it and at some point we're going to have to understand it and we're going to have to be able to calculate that.

John Furrier

>> You mentioned ITAM before, one of the big themes coming out of here, these interdisciplinary and also adjacent areas, departments, people, workers, converging together. What's been the big takeaway this week because it's been very collaborative in the discussion, but is it truly collaborative?

Robbie Plourde

>> No. And again, everyone's talking about it, but very few companies are actually doing it. There's two approaches that you can have. One is you have a role that is like a technologist that covers both. You don't see that a lot. The more interesting one or the one that's used more often is a technology council where both teams come together and work together and have that-

Paul Nashawaty

>> Arbitration.

Robbie Plourde

>> Exactly, because there is no collaboration. It's sad because the two disciplines really work well together when it's orchestrated properly, but they have different needs and they fall into different groups. You have some falling into procurement and the other is in the IT department, and they have very different needs and they usually don't work well together, finance and then FinOps bring those two together, but it's still not there.

John Furrier

>> We've been talking about this all week, and not laughing, but smiling and grinning and reminiscing. In cloud, remember shift left, security department and cloud, we're not getting along exactly either.

Robbie Plourde

>> Exactly.

John Furrier

>> Remember those days?

Robbie Plourde

>> Yeah.

John Furrier

>> And then all of a sudden, okay, shift left. Now you're in the CS. The developers, engineers are in the pipeline, great, solve. Is there a moment coming where you see maybe that kind of play to keep the no-ops saying no?

Robbie Plourde

>> I hope so. I hope so because what I do like about FinOps and the idea of tokenomics is it's a C level problem, which I'm hoping then trickles down to do this. In the past, it's always been a bottom up, like move to cloud. No one understood that. So the engineers were like, "Yeah, let's do it," and C-level-

John Furrier

>> Well, they loved it.

Robbie Plourde

>> Exactly. So hopefully now that it's C-level is understanding because there's costs associated to it, and it's an increased cost, not a reduction in cost, that is where I think it might change that approach a little bit, hopefully.

Paul Nashawaty

>> Yeah, that makes sense. I think that the approach is being evaluated and constantly reviewed. The pressures to use AI to help accelerate is always in place, but wait, where can things go very wrong with applying AI in FinOps?

Robbie Plourde

>> Yeah. It's really the speed that it's going and not understanding without the guardrails, right? Like I said, if you don't have that foundational data layer, AI is going to give you confidently wrong answers and you don't know... If you don't have that human element to review that or do the checks on that, it's going to get out of control, and we see it a lot. I think we're going to start to see companies realize that they've gone too fast and too far and it won't be a problem until we have something like something happens at a very large scale and everyone's going to be like, "Wait a second."

Paul Nashawaty

>> Well, doesn't that already happen though with the cost, the runaway cost of token?

Robbie Plourde

>> Yeah, but for some reason, people... If you go to Europe, it's a very different approach to AI than here.

John Furrier

>> How?

Robbie Plourde

>> They're trying to put guardrails and governance around it and they're taking a very paused approach. And here in the US, we love just pumping things through because of the productivity and the idea of it's going to solve a problem.

Paul Nashawaty

>> Well, double click on that because that's important. Last week, Europe put out the data sovereignty plan there. That went live. That's now in motion. The EU CRA is going into effect for applications. Reporting has to be done by September this year. All applications have to comply by December 2027.

Robbie Plourde

>> They're probably going to move that.

Paul Nashawaty

>> Well, I don't know. I mean, but there's severe penalties, including jail time for people that put their name on these things. So these are really, really strong penalties. Now, keep in mind, just because it's happening in Europe, doesn't mean it's not a global issue.

Robbie Plourde

>> I agree.

Paul Nashawaty

>> Because if you're running an application that touches Europe, it has to comply. If you're running an application in Europe that touches outside of Europe, it has to comply.

Robbie Plourde

>> Yeah. So you're going to have that cross at all. Yeah. I agree.

Paul Nashawaty

>> So there's a lot of things happening.

Robbie Plourde

>> I hope we adapt some of those.

Paul Nashawaty

>> Oh, I think we have to.

Robbie Plourde

>> We have to. Exactly. Exactly. Now, will that stifle some of the advancements we're seeing? Who knows? I'm hoping that we do get some guardrail or some governance at least.

Paul Nashawaty

>> Well, yeah, I think that's a good question. I think we were talking about this earlier. I think the question around the dollar is not equal to the dollar, meaning if you're innovating, that should be a little bit of a sandbox environment and the rules don't apply. If you're in production, all things have to apply, right?

Robbie Plourde

>> Exactly.

Paul Nashawaty

>> And so that's where the cost is, but there needs to be a sandbox that it's a safe zone for people to work in and not run up a huge bill.

Robbie Plourde

>> No. Well, if you've seen, they don't talk about the cost of it. They talk about the value of it. The conversation is shifted away from how much does it cost to what's the business value?

Paul Nashawaty

>> Yeah, but I agree with that, but the business value for production workloads and the business value for innovation, you can innovate and try things and if it doesn't work, okay, that's fine.

Robbie Plourde

>> What is it, 95% of something of AI projects are failing right now, something along that line?

Paul Nashawaty

>> I think, yeah, the MIT study is a little bit of a clickbait to me.

Robbie Plourde

>> Yeah, I agree.

Paul Nashawaty

>> This is coming from somebody from Sloan, so I get it.

John Furrier

>> And by the way, there's another half that's winning. Governance is a big role in all this. We're seeing success is the people who understand the data, to your point, I think that's where we're seeing the successful use cases. And here, I think is really more of an engineering culture, but it's driving up to the C-suite. So it's really deep tech meets C-suite. Now, but they're not the customer, but they're the beneficiaries-

Robbie Plourde

>> Correct....

John Furrier

>> if it because real time, unified language with focus. These are good signs. So with that, what do you think the future of this community is? Obviously is getting bigger. We're in migration.

Robbie Plourde

>> Exactly. And that's the practitioner's job is to calculate what that business value is so the C-level understands what they're doing with this AI and how it's being applied because we're learning new use cases every day. And as new models come out and you start to decide, "Hey, let's use it for testing. Let's use it for development. Let's use it for analysis," there's so many different use cases, and some are better than others, right? The whole AI to give you recommendations is great, but if you do that without the business context, it's useless because it's creating a recommendation. And sometimes, when they automate that and it actually turns something off without knowing the business context of, "Wait a second, that's a failover device anyways. Why are you turning it off? It's doing its job," you have to bring that business context into it.

John Furrier

>> All these adjacent costs that we're seeing, these hidden costs under the iceberg. Robbie, great to have you on. Great insight.

Robbie Plourde

>> Absolutely. Thank you.

John Furrier

>> Final question. What are you optimizing for this year? What's your focus? As this landscape evolves, you're in the front lines, you're a frontier player.

Robbie Plourde

>> It's really to understand that token, the tokenomics, and how can you apply the consumption based approach to the ITAM Foundation that understands the data, the business value and how to do that. So we're really focusing on getting that consumption-based analysis to help drive the use of it in AI.

John Furrier

>> Well, we really appreciate the insights. Love the energy.

Robbie Plourde

>> Thank you.

John Furrier

>> Given us-

Robbie Plourde

>> Appreciate it.

John Furrier

>> End of day two.

Robbie Plourde

>> Absolutely.

John Furrier

>> Thank you very much for coming on.

Robbie Plourde

>> Absolutely. Thanks, guys. Appreciate it.

John Furrier

>> I'm John Furrier, Paul Nashawaty. Doing our part to bring the energy as we bring up FinOp. A lot of energy here. A lot of inmigration into this community is going to be super relevant, is super relevant and it's going to continue to grow because when you track the money and the costs, match it to the value, that's where it's all coming together. Teams are being reformed. Things are looking good. We're doing our part here in theCUBE. Thanks for watching.