CloudHealth Roundtable: Leadership Responds to the State of FinOps 2026

On Thursday, February 19th, 2026, the FinOps Foundation released its 6th annual State of FinOps report, providing invaluable insights into the trends, challenges, and maturity of the FinOps discipline. In this blog, we share the discussion from our recent CloudHealth roundtable discussion where the leadership team reacted to the key findings of the report, offering their expert commentary and perspective on the current state of FinOps.

Before the roundtable on Wednesday, February 25th, Cathal Cleary outlined CloudHealth's response to the trends and challenges highlighted in this year's report. He discussed how CloudHealth's current features address these themes and also previewed exciting, purpose-built roadmap items designed to tackle them.

About the State of FinOps Roundtable

The CloudHealth leadership roundtable provided an expert reaction to the State of FinOps 2026 report, diving into the shifting demographics of respondents, the growing focus on AI, and the expanding scope of FinOps to include SaaS/licensing costs, private cloud and data center, and even labor costs. Some key topics of the roundtable included the sustained prioritization of waste reduction and optimization, how organizations are addressing AI initiatives, and the FinOps Foundation's decision to broaden its mission statement from "cloud and technology" to simply "technology.” Watch the full roundtable discussion for deeper insights into these critical industry trends.

About the CloudHealth Roundtable Panelists:

• Cathal Cleary: Product Leader, CloudHealth | Governing Board, FinOps Foundation
• Sanjna Srivatsa: Product Manager & Data Scientist, CloudHealth | FOCUS Contributor, FinOps Foundation
• Lucas Paratore: Product Manager, CloudHealth | Technical Advisory Council, FinOps Foundation
• TJ March: Product Marketing Engineer, CloudHealth
  
  

TJ March: Good morning, good afternoon, everyone. Thank you for joining the CloudHealth Roundtable. This is our first one of 2026, and today we're discussing the State of FinOps 2026 report, released last week by the FinOps Foundation. 

With us today, we have Cathal Cleary, our product leader and a governing board member at the FinOps Foundation. Sanjna Srivatsa, a data scientist, product manager, and FOCUS contributor with the FinOps Foundation. Lucas Paratore, a self-proclaimed FinOps nerd, product manager, and Technical Advisory Council member.

Let's dive into some of the behind-the-data stuff, starting with the respondents—who responded, what size of organization, and where in the world they responded. Cathal, we’re going to start with you.

Source: FinOps Foundation | https://data.finops.org/

Cathal Cleary: I'll get it going. Honestly, when we looked at the data, we thought there were really no surprises, maybe a little boring. AI takes over the world—that's a narrative we're used to. But we did want to look at who's responding. The number of respondents went up by about 100, an 11% increase, which isn't entirely in keeping with the growth of the community and FinOps itself; it's quite a bit lower. For a community this big, less than a thousand respondents is a little disappointing, and I think we need to think of ways to get more people to pitch in.

Looking at the regions, Europe and EMEA took over the mantle of leaders from North America. Last year, EMEA was 26%, now it's 35%. North America was 42%, now down to 34%. So, certainly, EMEA is beginning to overtake North America in response to this survey. Asia also pipped South and Central America, going from 10% to 15%. Shout out to everybody who's not in North America.

TJ: Let's keep going with some more behind the data stuff. The enterprises that responded—it seems like we have a great group of large enterprises. Cathal, I know you have some big thoughts on our demographic of respondents.

Source: FinOps Foundation | https://data.finops.org/

Cathal: I definitely think the responses are heavily influenced by the large enterprise. Large enterprise is 10,000 or more employees. So, I think all the data needs to be viewed through the lens of those super-large companies. Take GPUs, for example. Are SMBs managing GPUs? Probably not. Enterprise maybe, particularly if they're renting GPUs from the cloud. The answers to questions like that are definitely influenced by the fact that the respondents are, by and large, these large, large enterprises.

Sanjna Srivatsa: I do want to say that every enterprise, depending on its size, just behaves so differently. So, I do think that there might have been some overshadowing of patterns based on the participation, but they are fundamentally different in the way they work.

Cathal: Let’s move us to the next slide that kind of breaks down a little bit by these different demographics

Source: FinOps Foundation | https://data.finops.org/

The one thing that came up here is AI for FinOps. It was like number 3 for large enterprise in the future and like 2 for SMB, which is surprising it didn't come up in Enterprise at all. The way I see it is that people in the industry are looking to AI to make their jobs easier, or at least to make their jobs more efficient. Our partners are looking for ways to do more with less. They are under a super amount of pressure from a margin perspective, and there are changes in the industry. So, our partners really need to find ways to be more efficient. I think, like everybody else, they're looking to AI to do that for them. How can AI be my co-pilot for FinOps?

Lucas Paratore: On the flip side, is AI for FinOps more important to you as an organization? Because right here, across all three, FinOps for AI was the higher priority. Do people feel similar with that?

Cathal: What we've actually seen this year, in a lot of our customers that we've spoken to, is that they don't necessarily see the whole issue of spending on AI as a problem yet. This slide kind of reflects that, right? Everybody's saying it's a future priority. We think this is going to be a problem in six months' time, but it's not a problem yet. A few of them did come to us and say, I definitely need a good AI system to help me manage my FinOps in the meantime, so I want you guys to prioritize that. I want you to do AI for FinOps first. Thinking about these things as a future concern—everybody seems to be worried about the cost of AI, and I think that's reasonable because we don't necessarily have a good handle on that yet. We don't necessarily understand yet whether the cost of AI is going to be a huge deal in six months or not. But certainly, currently, AI spending does not seem to be one of the top concerns for any of these segments.

Sanjna: I really like that the organizational priorities reflect reality, right? Day-to-day, what is it that FinOps practitioners care about? It is the same base reduction, making sure we're clean, optimizations done, allocations done. But since I'm sure everyone's manager is kind of asking them to say, "Hey, what is this whole AI stuff? Can we get in on it?" that's why people are kind of investigating it. Unless you have significant costs in AI, it's not something you will spend your everyday on. It is something that you want to dabble in, kind of like what FOCUS was two years ago. It was the big new thing, everyone wants in, but no one really knows how or why it's valuable to you. I feel like that's kind of the hype cycle we might be in, so everyone kind of expects that we will make this a top priority in the future, but today our bread and butter seems to be allocation and forecasting and good old optimization and idle waste.

Cathal: Waste reduction is at the top of all three, right? It's still a thing.

Is AI for FinOps mature enough yet to trust it to go do, like, destructive things to your infrastructure? Our experience with automation over the years has been that adoption has been pretty low, and I think that probably is that trust problem. Do we trust the system to do those things on your behalf? The question is, as AI booms and blooms and grows and becomes more and more clever, will that trust build to the extent where you will actually allow AI agents to go do things in your environment?

TJ: Sanjna, as our data scientist, what do you think?

Sanjna: The way I read that question was actually that there's maturity required within the practice to handle automation. I completely agree with that thought process, that if you have all the guardrails and governance and all of that in place, you actually can allow for some automation, if it's AI or otherwise. Personally, I think AI is absolutely there, where you would, if you have all the guardrails and governance and processes in place, allow for some automation with AI. I don't see why not. Ideal waste reduction and stuff like that, that's absolutely something that I would personally think that we are ready for AI to do. RI management, maybe, maybe not, because that's far more incriminating on your bank balance, but idle waste, idle stuff, waste reduction, possibly even workload optimization with a human in the loop, I would absolutely think we're there.

Cathal: Yeah, I think that's the key, isn't it? We still have the human in the loop, so it's not complete trust, it's trust but verify. The AI is not making the decision, it's providing decision support so that that decision can be made quicker. No real surprises here. I think it's interesting to see everybody is still prioritizing workload optimization, waste reduction, and I think this reflects that the AI initiatives need to be self-funding. It looks like people are still pushing hard on optimization and waste reduction to self-fund the AI initiatives. FinOps is showing its value to the organization by providing that budget relief so that we can actually invest in these AI areas.

Sanjna: That is so interesting that you said that. We're looking at AI as this completely separate entity, right? We would never kind of take optimization money and fund it somewhere else if it was just a compute spend. This is not how the cycle works. But we're treating AI as this special category. We'll do something, and maybe the result will pay off by many folds, almost like gambling. I feel like a lot of us are going through the cycle where we are not starting with the use case. Unless we start with a use case to say, "This is what we want to solve," and you know exactly what we're building, and that's why we will use AI, you'd stop looking at it as this money you're spending into this black box, but money you're strategically spending into something you exactly want to achieve. I think a lot of us enterprises are going through that cycle right now. It happens very often with hypes. It's a little interesting to observe that we're treating AI as this completely separate thing when all that it is, is just another service that CSPs provide. It's nothing different, right? Why would it be?

TJ: Diving into a little bit about the numbers that we saw at State of FinOps: What are the areas of technology spend that FinOps teams manage today and in the next 12 months? This is one that State of FinOps releases every year. Lucas, I know you have some thoughts on this.

Source: FinOps Foundation | https://data.finops.org/

Lucas: I was a little surprised by some of the data here. Especially, I mean, from the bottom up, the 28% beginning to include labor costs. I was really surprised by that. Maybe it's an outcome of FinOps teams moving under CIO/CTO, so now they have more access to labor data for the engineering organization, but that one really didn't make too much sense to me. I'd be interested if there are other people here that have been thinking about how they're going to take labor data to do things like calculate unit economics, understand profit margin, or another use case. We're definitely seeing that combination, those personas, combining ITFM, TBM, FinOps.

Sanjna: Does anyone else think it's surprising that AI, while AI might be the same as licensing or SaaS, depending on how you're using AI in your company, or data center too, does anyone find it interesting that it is standing as its own pillar? Because most of us consume AI basically in any of these other formats that are also listed on this page. But it is standing on its own with a really huge number behind it. Do we think it's interesting? Do we also consume AI separately? Are we tracking those costs separately? Or are we consuming them as a part of SaaS license or public cloud costs?

Cathal: That's the real question, Sanjna. How is AI not part of SaaS? How is AI not part of licensing, right? But AI is also public cloud, and obviously, this is the areas outside of public cloud. But AI is public cloud, and there was a bit of debate in the FinOps organization, too, about: Is AI not just a cross-cutting concern, or is it a thing on its own? It looks like the number was so big this year that they decided to make it a thing on its own, but last year, this didn't appear on the data that they shared. Even though there's, like, a 35% year-over-year change, when I looked at last year's same group of numbers, it was just SaaS licensing, private cloud, and data center.

TJ: One thing that is not on this slide, Lucas and Cathal, do we want to talk about one of the announcements that was made on the State of FinOps call about the mission statement change?

Cathal: Governing Board plus TAC, I'll let Lucas talk to TAC, but the governing board decides what the mission of the FinOps Foundation is. The mission previously had said, "managing the value of cloud and technology," and I think we just felt that cloud is technology, and why are we calling both out? Why is cloud special? The purview of the FinOps persona is now much wider than cloud. We've got all these things that you see on the slide here, so why are we calling out cloud special? It was a little bit of back and forth, and people felt passionately that FinOps had grown up in the cloud. But ultimately, we thought, look, the mission statement absolutely needs to be clear and simple. FinOps is in, now and into the future, going to be managing the value of technology. So we changed the mission statement, we took cloud out, and we just said, technology is the value we're managing. That's where we're at.

Lucas: The Technical Advisory Council, we have not yet made that change. We still refer to it as cloud and technology, but TBD, we will probably be moving in that direction as well, but we also feel the same tie and affinity to the word cloud.

Cathal: To be clear, the difference here is the mission of the FinOps Foundation versus the definition of FinOps, and I think, Lucas, that's what the TAC is deciding on "what's the definition of FinOps", and should that change?

Lucas: Yeah. Exactly.

Sanjna: At FOCUS, as a contributor, we are constantly being pushed towards also considering SaaS use cases when we write our specs. Which is very interesting. We were most recently handling commitment management, and the way we define commitment as a cloud provider is very different than a SaaS provider, and so now we're actively seeing that our scopes are expanding, and we have to consider how the others do things as well.

TJ: How have priorities changed? Staffing year over year is still the bottom of this chart, same as last year. Multi-investments are starting to show up as a priority. Unit economics still seems like it's a little low, but then we also see a big push from the FinOps Foundation in leadership and executive integration or influence, but that also seems low. Cathal, you had some other thoughts.

Source: FinOps Foundation | https://data.finops.org/

Cathal: I'd like to start at the bottom, because I think the top, there's no surprises, really, at the top, but staffing and acquiring talent has been at the bottom for the last two years, and I was wondering, why is that? Do people feel like their teams are fully staffed, or do they feel like there's so much training out there for FinOps these days, we don't really have a talent problem anymore, people can get sufficient training, or do people feel like I don't have budget to hire people, and I have all this AI cost coming in that I need to find money first, so I'm really going to lean on AI to help me do more with less, kind of the point from earlier. Why is staffing and talent such a low concern?

Sanjna: Have you seen any companies mandating that we use AI where we can and not hire any more people? Is that a thing at all? I believe it, right? I see how we might do that, too. We might hire less engineering, we might have more generalists now, people who can do multiple things. I see that being the world a couple of years from now, for sure.

Cathal: Before you come to me with a requisition, prove to me that AI can't do this job, is that kind of where we're at? I kind of feel like even in recent earnings announcements for public companies, it's been cool to talk so much about what you're doing with AI, and how much or how less you're hiring because of AI. I mean, it almost seems like if you say that in your earnings review, you'll probably get a share price bump, which is a little unfortunate.

Sanjna: I see some really interesting conversation happening in the chat. We're talking about how we're categorizing services. Five years ago, everything was ML, right? Like, what do we use for the most basic linear regression? ML. Nothing was statistics, nothing was below, everything was ML, and now everything is AI, so I see how it makes sense for a lot of these things during the transitionary period when we move between ML and AI for service category to kind of consider more ML services as well.

Cathal: I think even last year on this particular slide, it was called AI/ML spending, right? Or something like that? It wasn't FinOps for AI.

Sanjna: ML was the most fun thing, everyone wanted to talk about it. Even if you did something on Excel, we called it ML.

Lucas: Something that's interesting, too, to call out here that stayed consistent is the implementing a new FinOps tool. So, maybe a sign of what the market is, is that whether that's a spot solution or not, organizations continue to invest in FinOps tools at pretty much the same rate they have been the last few years.

Cathal: Are we seeing more build your own at this stage, or are we seeing more people actually turn to cool vendors like ourselves?

Lucas: I think it's company-specific. There's quite a few characteristics of a company that drive one versus the other. I think we are seeing some more build happening as cloud providers increase their free tools, but I also think that organizations are becoming more complicated, and so as you take over multiple clouds, you have private cloud, you have SaaS, it's hard to lean on the investments as much that the cloud providers have made, because they certainly have made some good progress. So, as the world becomes more complicated, I think there's... it goes kind of both ways.

Sanjna: I've also seen build on the stuff that you buy, not just build and buy, but, like, build with the stuff that you buy. A lot of people are taking what we have and make it their own, or taking what CSP providers provide, and then they make it their own. Seeing a lot of that too, and it's easier to build now with AI, right? Everyone can be a builder now, so I'm seeing some of that as well.

Lucas: That's been the MSP model since the beginning, is how do you build upon... they don't just go and take a CloudHealth or whatever and say, "You're getting 100% of this platform," they sell an additional service on top of that. So it's how do you build upon these tools to provide a better product. 

Cathal: For unit economics, even the language for unit economics changed, right? Last year, it was "getting to unit economics," this year it's "defining unit economics." I don't know the difference, honestly, or why those terms changed. The unit economics indicator, which was in place number 4 last year, is actually in 7 this year, so it's dropped 3. As a SaaS product that has to measure our revenue carefully, we do certainly look at unit economics for CloudHealth, and we look at for every dollar that we manage, how much does it cost us to manage that dollar? Everything we do revolves around that unit economic measure. There is a finite cost every time we have to rerun some of our pipeline or some of our jobs, there's a finite cost that we have to be aware of and we have to measure and make sure we're allowing for.

Sanjna: I was just gonna say how, when we have our prioritization product feature meetings, the one... if I could productize one thing that I know is the most impactful thing is executive influence. If there was any way to productize that or make that into a feature, we've first-hand seen how important it could be. It is by far the single most impactful thing on someone's cost. If your manager or if your CEO says you have to do something, you will go do it, and it will just happen. Things that you never thought—if you had to cut 50% of your cost, you're just like, "Are you crazy?" But then, if your CEO says you have to do it, we kind of see that happen, and so we kind of have this running joke. If there was any way for us to productize this kind of influence, we absolutely would, and that would be the best thing for a practice.

TJ: Sanjna, you've hit on FOCUS a good amount already. Is there a lot of asks around getting FOCUS-compliant data into data center at the working group? Is that a big focus on FOCUS?

Source: FinOps Foundation | https://data.finops.org/

Sanjna: I think we've reached a point where we have a lot of voices, so the spec is molded by the people in it, right? The contributors. If it's fair or not is a separate question. We try to make sure we consider everyone's opinion, but we can't without those people actively there. For example, just the definition of "effective cost" took us 25 hours plus, just to discuss it. The FOCUS community is a very active, everything takes work, everything takes time, everyone's voice is heard, kind of a community. We don't have nearly enough voices of companies that use data center and private cloud very actively. So I feel... right now, we have a lot of SaaS voices, so I think that's showing in the spec. Public cloud, obviously, is something we have, some licensing voices as well, but I know we need to support data center costs, and we do actively try to consider everything, but since we don't have someone telling us all the constraints and all the scenarios, I don't know if we do the best job that we can. I guess we have to, but we just don't know how we do it at this point. This is something that has to happen in the future.

Cathal: I think the interesting point going forward is, like, what are people gonna do now, right? You've seen the data. Are you making any changes? Or is it just like, yeah, we kind of expected all this, no problem? Everything is gonna stay as is, and our plans are gonna stay as is? Obviously, the goal of a survey is to influence, you know, maybe your actions in future, and to inform you maybe a little bit about what the future holds. So, has anybody got any plans to make big changes based on this data?

We started by saying the results were a little boring. That's actually a good thing. I'm glad that we didn't have any huge surprises, it means that we're doing our jobs.

TJ: As we're winding down our first Roundtable of 2026, I want to encourage everyone to continue the conversation. Join us in the CloudHealth Community  that has been created for CloudHealth users to share best practices, share our blogs, but also it is a forum, so if there are things that stuck out to you as you are diving into the State of FinOps data a little more internally, would love to continue the conversation in our CloudHealth community, and as peers and FinOps practitioners in the wild, so please feel free to join us.

Cathal: Just a couple of other quick comments as we close. We purposely avoided telling you guys anything about CloudHealth and what CloudHealth's doing in response to this data. There is a blog out there that I wrote that kind of talks about this data in the survey, and what we're doing, and what we plan to do, so that is a blog in this community.