FinOps in Action

What if the biggest driver of cloud waste isn't what shows up in your billing dashboard?

In this episode of FinOps in Action, Taylor sits down with Mike Jaco, who spent years at Mastercard running one of the most sophisticated hybrid cost programs in enterprise technology. A unified TBM and FinOps practice covering a multi-billion dollar technology budget, on-prem and public cloud, under a single common taxonomy.

Mike's core belief: most large organizations will be hybrid forever, and the industry hasn't caught up to what that actually requires.

Here's what they covered:
  • Build a common taxonomy before you do anything else. If your technology teams, product teams, and finance teams are using different languages to describe the same thing, no one is having the right conversation. Get everyone onto one shared structure first and everything else gets easier.
  • Allocate cost down to the product level, not just the business unit. Stopping at the business unit level leaves too much room for ambiguity and not enough room for accountability. The more granular your allocation, the clearer it becomes who owns what and what action to take.
  • Most large organizations will be hybrid forever, so build for it. Regulation, data sovereignty, and steady state workloads mean a full cloud migration is not realistic for most enterprises. Build your measurement framework to cover both worlds under one taxonomy or you are only ever telling half the story.
  • Savings that stay in the team drive more action than savings that disappear. If engineers know that any efficiency they find gets reinvested into their own backlog rather than returned to a central budget, the motivation to optimize shifts completely. Self-funding growth is a more powerful incentive than a cost reduction target.
  • FinOps is a people problem first and a data problem second. When an engineer is not making a change, it is almost never because they want to waste money. They are overloaded, they have competing priorities, or the task does not benefit them. Solve for the people dynamic and the technical changes follow.
Chapters:
00:11 Meet Mike Jaco
01:11 Why Hybrid Is Forever
01:38 Regulation Drives Reality
02:55 On Prem vs Cloud Economics
03:28 The Apples to Apples Problem
04:58 Building a Common Taxonomy
06:21 Making Tech Less a Black Box
06:59 Why Allocate to Product Level
07:13 Accountability Creates Action
08:36 TBM vs FinOps Actions
10:21 18 Months vs 3 Months
11:45 Granularity vs Peanut Butter
13:16 Controllable vs Uncontrollable Spend
14:45 Carbon Becomes a Currency
17:11 Cloud Carbon Data Gap
20:39 Cost vs Carbon Tradeoffs
29:54 AI Spend and FinOps Future
37:39 Career Next Steps and Advice
39:47 People First FinOps Lessons
41:22 Where to Find Mike

🎧 Subscribe for more FinOps leadership conversations → https://www.finopsinaction.com/ 

Quote of the Show:
  • “You can do more with the same budget.” - Mike Jaco 
Links:
Ways to Tune In:

Creators and Guests

Host
Taylor Houck
Producer
Serena Johnson

What is FinOps in Action?

Welcome to FinOps in Action! Join host, Taylor Houck, Each week, as he sits down with FinOps experts to explore the toughest challenges between FinOps and Engineering. This show is brought to you by PointFive - empowering teams to optimize cloud costs with deep detection and remediation tools that drive action.

FIA - Mike Jaco
===

Mike Jaco: [00:00:00] the other thing I would say, especially early in, your career, is you can't do everything. You know, right? You're, eager to prove yourself. You're eager to work and stuff like that, but you have to find the balance, and you have to find the trade-offs. and you know, I think that's where something that I learned is sometimes it's more important to talk about what you're not gonna do than it is what you're going to do, right? 'Cause prioritization is hard. Everybody wants everything, you know, right? And having that tough conversation and saying, "We can't have everything. Here's the trade-offs. Here's

Intro: welcome to FinOps in Action. I'm your host, Taylor Houck. Each week I'll sit down with FinOps experts to explore the toughest challenges between FinOps and engineering. This show is brought to you by 0.5, empowering teams to optimize cloud costs with deep detection and remediation tools that actually drive action.

Taylor Houck: Hello, and welcome to another episode of FinOps in Action. Today's guest spent the last several years running a program that most FinOps leaders never get to [00:01:00] run. At Mastercard, he led a unified TBM and FinOps practice across a multi-billion dollar technology budget, covering both the on-prem footprint and the public cloud under a single common taxonomy.

By the end, his program had directly allocated roughly seventy percent of that spend down to the product level and built a carbon allocation model that ran in parallel to cost. Before that, he spent years inside large enterprise technology and infrastructure organizations building the muscle for cost discipline at scale.

He believes that most large companies will be hybrid forever, and the real unlock for FinOps is getting both worlds onto the same measurement framework so leadership can actually make decisions. Welcome to the show, Mike Jaco

Mike Jaco: Thanks, Taylor. Really excited to be here

Taylor Houck: Yeah, man. Thank you so much for coming on the show, and I'm really excited to have this conversation [00:02:00] with you. And one of the main reasons is because of your experience operating in this hybrid, this multi-cloud and on-prem world. Because, you know, a few years ago, the whole conversation was move everything to the cloud.

You came out of MasterCard with a, a different view that most organizations are gonna be hybrid forever. Can you walk me through how you landed there?

Mike Jaco: honestly, we were-- I was a little bit lucky. So MasterCard is highly regulated, as a lot of industries are. Um, we're considered national critical infrastructure in a lot of countries now, and so regulation has required us to, um, go on soil in a few different places. And from the signals, it looks like that will only increase.

And so when I first, um, got into FinOps, went to the first FinOps conference, uh, it was twenty twenty-three, um, it was that feeling that everything should move to the cloud. And for one, that didn't sit right, but also I knew that we couldn't do that from a MasterCard perspective because of regulation. [00:03:00] So we were kind of lucky in, in a little bit that way. Um, but that's the way that we always started with FinOps, is that we will always be hybrid. And I think as time has gone on now and the conversation has shifted, you know, in the last, you know, several years, is that it just doesn't make sense to run everything in the cloud. So I think, you know, there's a lot of reasons why you might be fully cloud and, you know, in small companies, a startup and, and stuff like that.

But as you get into larger companies, you get it into regulation, data privacy even, um, then you're most likely going to have a hybrid, you know, type environment. Um, and I think that with the hybrid environment, you know, FinOps itself only tells part of the story when you're in a hybrid environment

Taylor Houck: It's interesting, and I actually have heard the same sentiment from others, right? And the general idea that I've heard, and I kind of want to hear your, your feedback on it, is that, hey, if you have very steady state applications, you're often better off running them on-prem because it's like, you know, why would you rent a home that you could own if you know you're gonna [00:04:00] need it forever?

Is that part of the, uh, the discussion?

Mike Jaco: Yeah. You know, I, I think it definitely is. I think that's the starting point. I think as you get into larger and larger decisions, there's so many factors, you know, that come into that, and that's the real challenging part. Um, so when you think about, you know, FinOps in the public cloud, you get everything in one unit cost, you know, right?

That is everything from your, you know, your patching, your upgrades, your refreshes, um, you know, even support structure in terms of, uh, sourcing, finance, and all of those functions and capabilities, those are all wrapped up into one, you know, price. And so, you know, when you're looking at making decisions, how do you properly make that decision to go from, you know, do I want to put this on-prem or the cloud? You're talking about some, some early data in terms of the workload itself, but most decisions are much beyond just the workload level. You know, when you're making it, am I gonna be in the, uh, uh, hybrid environments, um, and which one, know? And so I think that [00:05:00] that is a, a big piece to it is, is trying to get to an apples and apples comparison when you're actually comparing apples and oranges.

Taylor Houck: Yeah. It's interesting, right? Because in the cloud it's, it's even though, let's say in AWS terms, the CUR is very complex and large and perhaps not intuitive to understand, all the cost is there.

Mike Jaco: Mm-hmm.

Taylor Houck: The on-prem world, there are all these other expenses that come into it, right? Of course you have the hardware, you also have the land, you have the people, you have the maintenance, everything.

So can you walk me through how you think about getting to understand the total cost of ownership in a hybrid environment?

Mike Jaco: Yeah. Um, so I think, you know, one of the, you know, first things is this thing that we called building a common taxonomy. And so the first thing that we noticed is that we-- You know, early in my career, I sat in the intersection of technology, um, the product business [00:06:00] side, and finance mostly, right? It's the three, you know, kind of pillars. And they just could not talk to each other. They could not talk the same language. What the business side, you know, described as a product was much different than what technology call-- you know, that supported that product, you know, right? And finance, you know, kinda stuck there trying to figure it out and make decisions. Nobody understood really each other, and so they, they weren't having the same conversation. And so we had to build a, a, a taxonomy, and it started on the technology side, you know. So okay, so on the technology side, how are we going to map to the products that, that we provide? Ultimately, it's the market-facing products that we started with and, and, and, you know, what technology is supporting that.

But then we took that same, you know, mindset and methodology, you know, internally, and everything was, um, a, a product. So internal products, uh, market-facing products, we treated them all the same. And being able to to them in the same language, be able to, um, well, ultimately, you know, kinda tagging is kinda [00:07:00] how you, how you get there, but really being able to have that same conversation and bring the visibility into it.

And I think that's the first step in everything. And as you talk about, um, whether you talk about FinOps or you talk about TBM, the first step and the biggest challenge is always that, that initial visibility. Because things get kind of built on top of each other and on top of each other, and it's technology can become like a black box to the rest of the, uh, of the organization, of the business, not knowing kind of what exactly supports what and, and how much it's costing to run certain things, you know, you know, that.

And, and so the right conversations just weren't happening. And so, you know, kinda took, had to take that step back and, and what do we need to do? And, and first it was building that, that common structure, that common language, that taxonomy that we could all kind of center around.

Taylor Houck: How did you come to the decision to allocate down to the product level? I, I, I know that a lot of folks and a lot of programs, they kind of stop when they get to perhaps the business unit level. What, what pushed you to, to go that next step further?

Mike Jaco: Uh, [00:08:00] honestly, it's a desire for accountability and action. So you can't action off of data that you don't have the granularity into, you know, right? And so it-- and, and when you don't have that granularity, you also don't have that accountability. So you can never fully get to, okay, you know, here we need to make some efficiencies here.

You know, we can play a lot of games when there's not like very clear, you know, data, and it can become more of a people problem, and there's a lot of churn that can go on with that and, and stuff. And so we wanted to get to accountability for one, you know, right? And so you are, are the responsible owner for this.

It is your responsibility. So we had product owners, you know, uh, on the te-technology side, uh, as well as on the, obviously on the, on the product side. We had, um, uh, program owners, you know, each level of our taxonomy and had owners and accountable people that were responsible. So you know who is exactly responsible for, for that thing.

And it was to drive the action, to drive the efficiency that we wanted to [00:09:00] see

Taylor Houck: And can you talk now about that action that you were looking for? I'm really curious as to whether you were looking to help people make the decision to run their application on-prem versus in a cloud world, or is it about maybe the next incremental, um, build or, or feature that they were working on?

What are those actions that you were looking to drive folks towards?

Mike Jaco: Yeah. And, and so I think this is where-- So you have to-- I think, I think there's actions at the top senior level, and there's actions at the bottom, you know, kinda granular level. TBM and the on-prem side of things was really more of a senior leadership focus, you know, to really be able to get that visibility into where our technology costs, you know, what was generating our technology costs, what was causing, you know, um, you know, us to hold onto so much data, us to have so many IP-- whi- whi- which exactly products, which exactly, you know, were it, so that leadership could make better decisions and run the business more efficiently, right?

And I, and I think that was the first, the, the first step was, was that for, for the [00:10:00] on-prem side of things. On the cloud side of things, because in, in a way it was more straightforward, um, in terms of, you know, you don't, don't necessarily have to go allocate all that messy of costs, you just have to tag your data.

And then because we had our taxonomy, so the taxonomy came from on-prem, you know, right? And, and that's where FinOps didn't have to create the taxonomy, we just had to align to the same taxonomy, um, uh, as, as kind of that was already there and, and created by the time, you know, kind of FinOps became a, a true function and discipline at Mastercard.

So they benefited, you know, from that. Um, and so, but it, but it was that same general concept, bringing the visibility to it. It's just that with the visibility that FinOps brought, you could get to actions quicker, right? Because you knew exactly who owned things, you know, who was running the pieces. And so we were just able to get to that action and that accountability quicker by also benefiting off of, you know, what we did on the on-prem side.

If we wouldn't have had that common taxonomy, we would have had nothing to, to tag to, and then it [00:11:00] would have been a lot of, you know, reaching and, and, and a lot of, you know, not being able to hold people accountable for being responsible for different parts,

Taylor Houck: Yeah, this is a really, this is a really important conversation, and I want to touch on a, a point that came up in our prep call, which is that it took you guys 18 months to map all of your on-prem costs. But when you started ingesting the cloud data, it only took you three months to get over ni- and, and got over 90% of the tags in the first year.

Can you talk about the contrast? But first, fact, fact-check me on that.

Mike Jaco: Yeah

Taylor Houck: and next, can you kind of put that in, in context,

um

Mike Jaco: You know, so, um, so that is correct. Um, I think for one, you've got to talk about the size, you know, of it. Um, you know, the on-prem footprint was, I was trying to do the, the math in my head, twenty times the footprint of the cloud, you know, right?

And so just the sheer size of, of that, you know, right, and, and MasterCard and their data centers and the, you know, thirty, forty years of, of just [00:12:00] building and building and building on top of each other and growing and expanding in size, there was a lot to work through there, you know, right?

Where the cloud footprint was, is much, much smaller and much, much more straight, you know, kind of forward where you can just kind of tag those workload as you're deploying them, you know, as they go out. You know, automation is a lot easier kind of in that case and, and everything. Whereas the, the on-prem side just took a lot more, you know, going in and digging in and analyzing and finding things and figuring out how do we allocate the databases?

How, how, you know, how do we use-- how do we allocate consumption-based, right? Because, because the goal was not to do peanut butter. That, that wasn't the goal. You know, the granular data to drive the action and hold people accountable, that was the goal. And so, you know, we had to get to that granular level of data, and it just took more time to work through it on-prem, and there-- it just was more complicated. And that seventy percent you mentioned, we got to seventy percent of, of cost. I'll just clarify that we, we allocated a hundred percent of cost. Um, we got down, we, we, we uh, [00:13:00] seventy percent of cost and consumption-based, so P times Q.

Other stuff, we just kind of did do the peanut butter spread, the executive leadership, some of the networking cost, right?

It just didn't-- the, the value wasn't there to get down to that kind of granular level. And so we made decisions on where we wanted to get that granular versus not, and we decided that that seventy percent was about where we focus on the granularity at, and then peanut butter spread the rest of it out.

Taylor Houck: Really interesting. What, what in that, uh, in the portion of spend that you were not able to directly allocate, you mentioned networking as a big piece of it, but what were the other big aspects or big portions of that, um, kind of non-directly attributable spend? And how did you determine kind of the...

'Cause I know you can say, "Hey, peanut butter spread," uh, you know, based on a weighted average of how much each product is spending on the other areas. Is that how you approached it or, or how do you think about that?

Mike Jaco: so, so we, we had the different levels, you know, right? So I'm talking about product level, um, that was where that is. You also have program [00:14:00] level, and then we had business unit, you know, kind of on top of that. So really where a lot of that was there was, um, you know, networking costs because w- the, the, the other goal, so we're, we're trying to drive action, you know, right? What action are we trying to drive? We're trying to drive owners to, to, to take ownership of their, their product, their service, you know, whatever they're responsible for, take ownership of that, run that most efficiently, support the business, bring value, you know, out of it, you know, right? Um, if we-- if, if they make a decision to, um, uh, uh, refactor something and it, and it saves some, uh, you know, percent, they might save some compute there, right?

They might be able to on the on-prem side, they might be able to turn down a few servers, but they're not-- we're not gonna be able to, to lower our network cost because of, of that change, you know, right? So they couldn't affect that change. We're trying to really focus on things that they could affect. Um, and, and them reducing some, some, some cap- some capacity or something somewhere isn't gonna affect our overall network, you know, kind of thing. Um, information [00:15:00] security was another, you know, kind of area. That's a g- everybody-- Y- you don't get a choice on whether or not you're going to be secure or not. So MasterCard you those costs. Those were not controllable, I guess, you know? And maybe that's the better way, controllable versus uncontrollable is, is probably the kind of way to see it.

We didn't want people trying to say, "Oh, maybe I can save some money here by, by reducing my information security resource." You know, "I don't need three of them. I only need one of them," you know, type of thing, you know, right? We-- It was the behaviors that we were trying to drive. That was the main reason for, for, for those

Taylor Houck: Really interesting. And now I, I do want to shift gears slightly because I know that at MasterCard you were also focused on a, a different angle than, than solely allocating the cost, and that was carbon. And carbon and sustainability is a topic that seems to be coming up more and more as the years go on.

So I want to talk a little bit about what happens when cost isn't the only currency that you're working on. Can you talk just first a little bit about [00:16:00] why you built a carbon allocation model and how it differed from your cost allocation?

Mike Jaco: well, MasterCard is, you know, takes sustainability very seriously. They have, um, twenty forty, uh, net zero goals, and they follow all the, you know, SPTI, um, uh, standards when they're reporting and doing everything. And they had already had a very strong tech sustainability team that was in the background doing similar work in terms of, you know, that we were doing from a cost and expense side on the sustainability side in terms of bringing in the sustainability data, the server data a-and all of that.

They were mapping where our, uh, electricity, where our water was coming from, and they were, you know, using the different resources. And they had come up with a really, really robust, uh, robust set of data for our on-prem stuff, you know, that we owned and, and we ran. And what, what happened is, is as we were building up the, the TBM and the on-prem model, you know, they, they just ki-- just conversation just kind [00:17:00] of happened is could we do something similar with this base data set?

It's a base data set very similar to our expense data set. And he said, "Yeah, we absolutely could." Um, and we partnered with them, and we were able to just almost mirror the carbon model. Um, the only thing that we didn't, you know, do is when we have multiple... So the difference would be when you have multiple internal shared layers, um, on the cost side, we're passing those, you know, internal product to internal product to internal product all the way till it gets to mar- to the market-facing product.

In the sustainability side, we kinda skipped some of those internal. We did direct consumption and then just where it ends up at the market-facing products. We kinda skipped some of those internal, you know, kind of transfers or whatever to make it a little bit more simple. But ultimately, it was able to guide us to where, you know, to see where these products, these regions, you know, these areas are where, you know, we're, um, generating the most, you know, carbon. Um, and so because MasterCard was ta- you know, taking sustainability so seriously. It had the on-prem data. This is where to kind of on the FinOps side, the, [00:18:00] the growth of the public cloud expense, you know, the sustainability team or our corporate sustainability officer came and said, "We want this same data that we already have on-prem for the public cloud." And so, you know, we went searching for that. And, um, it was a, a, you know, long run and I think it ended, it just shows the, um, the importance, uh, that MasterCard is, is kind of putting behind that because, um, you know, there was nothing really out there that gave us the granularity of data that we needed.

Um, you know, 'cause we were, you know, um, you know, granular at the product level, but also, you know, real-time enough to be, you know, accurate. Most of the things that, you know, publicly that we saw out there were months or years old, you know, kind of worth of, of data. And so they asked us to, to go searching in the market, and we were able to find, um, a company in the market that was able to enrich the public cloud data. Um, name of the company is Green Pixie if, if, if anybody's, you know, looking into that, and they are amazing and do some amazing work. Um, and they were able to enrich [00:19:00] all of our data, um, and, and basically just gave us the electricity, the water, um, the carbon, and then help us convert that all to a carbon equivalent.

So the MT, you know, CO2e, so we can kinda have one measure, right? There's one expense, one cost, you know, y- you know, USD, one expense, call it, from the, the sustainability side, and that's MT- CO2e. You know, and then, um, and then from, from that aspect, the-- we didn't even have to build a separate model. It just allocated, you know, based on the tagging.

Um, once we had the data, we were already allocating all that stuff from an expense perspective. Once we were able to enrich it with the sustainability data, it just flowed through the same, know, the same model. And then that just gives another to make decisions off of, you know, right? And so even if you're not, know, marching towards sustainability goals, you know, should be treating our, you know, what?

Earth resources the same way as we're treating our money, right? Just using it as efficiently as possible. [00:20:00] You can take out all the politics, all the climate change, and all that stuff out of it, and just why would you not be using it as efficiently as possible, you know, right? And this data just, um, gives you an extra piece of data to make better decisions. And it will help make better business decisions, you know, in the future, especially in hybrid environments and especially sustainable-- Sustainability becomes more and more of a thing. And I know at least the last that, that I remember, you know, hearing is that, you know, um, uh, the regulations in Europe are still kind of going forward, and so regulation is going to force it whether we want it or not

Taylor Houck: Oh, it's definitely coming and I'm hearing more and more about it. I'll tell you, there's a couple different directions that I, I wanna go in. First, I just wanna make a note. I'll tell you, at NBC Universal, sustainability became a, a big conversation for us in our FinOps program, and we went down a similar journey of looking for the cost-- uh, sorry, the, uh, the carbon and, um, sustainability data coming out of the native cloud provider tools, and also found that it was lacking.

But even taking a kind of a, a [00:21:00] finger in the wind type of a measure of it, I found it also really helped me drive engineering action in ways that sometimes cost didn't. Because as you mentioned, everyone cares about our planet, right? And no one's gonna go out there and tell you that it's better to use more water, more carbon versus less, right?

And some of the engineers that I worked with, when I went and talked to them about cost savings, it was like, "Hey, man, it's not my money," right? But it's everyone's Earth. Um, but here's a point that I wanna run by you because it seems like you've gone very deep on this topic specifically in a way that many others haven't, and it is this: I've heard so many times from people that because carbon allocation and measuring sustainability in the cloud is difficult, you can simply use cost as a proxy and say, "Hey, if you take an action that reduces your cost, you are likely reducing your carbon," because generally speaking, a portion [00:22:00] of the cost that you pay is for the energy that it takes to run this service.

I'm curious, when you got to that product-level allocation, did you find that there was a direct kind of, um, relation between the cost of a product and its sustainability or its carbon impact? Or did you find sometimes there were like high carbon s- users that were less high spenders, relatively speaking, within the organization?

Mike Jaco: in general, in-- when you're talking about actions, any, uh, most actions, I would say, you know, over ninety percent of the actions you're gonna take to save costs is most likely gonna save carbon. You know? And so in general, you know, if you don't have the investment, if you don't have the backing, you know, right, you know, to do it, I think that that is, uh, you know, could be a general rule of thumb. But, you know, you can't, can't, uh, you can't really make decisions off what you can't measure. So when you're-- You can't really measure it, you know, there. also more than just, okay, it's not just an efficiency, you know, thing, uh, especially when you're in [00:23:00] a hybrid environment. And now I'm getting more into the leadership, you know, you know, aspect of it. There, there's a regulatory aspect of it, but then there's the decision-making aspect of it. And there are obviously vast differences in the sustainability impact of different regions around the world where they're getting their, their, um, you know, energy from and stuff like that. And so even just where to deploy, you know, right?

Um, that, you know, can, can, can show that just deploying it in a certain location and cost being the same, you can save carbon. So there's a lot that you can do without saving costs, you know. So when performance, when latency, when stuff like that isn't an issue for you to necessarily stick something in a very specific location, I think if I remember right, it's been a few months, so don't quote me, but I think, you know, US West, you know, versus US East is dramatically different in terms of the, the, the, um, carbon impact and output that they have.

So if you have a choice to put in one versus the other and the costs are relatively the same, why not, you know, you know, do it that way?

Taylor Houck: Which one, [00:24:00] which one is better?

Mike Jaco: think US West was, was, was better than US East, and it was quite a bit Um, and so you'll see-- And but you also see the impact. So we were able to show that, you know, the amount of capacity in one versus the amount of capacity in, in the other, and then the carbon impact of those were not, you know, proportional.

Like it was, it was definitely, you know, higher proportions on, on the higher side. Um, would also say that there's always the opportunity to do more, um, as well. And so yes, making a, a decision to save some cost will likely save carbon. But you could also, you know, and what we showed is that, you know, you, you would-- you most likely will have three decisions in, in, in, you know, big, you know, types of decisions.

You can save just on cost, you know, and, and maybe not save any carbon, right? You can find the middle ground between saving on, on cost and carbon, or you can potentially spend more or save less to save more carbon. You know, right? And depending on where regulation, [00:25:00] what your goals are, if Mastercard wants to hit-- If any company wants to hit net zero goals, eventually they will have to start weighting carbon as a higher factor than cost in decision-making.

Taylor Houck: That's so interesting, and it really gets down to this kind of the, the bow on this entire conversation, which is that getting more data in the hands of your decision makers

Mike Jaco: Mm-hmm.

Taylor Houck: Help them in the sense of they're making this calculus of what they're optimizing for. And with a lack of data and a lack of information, you're making an uninformed decision.

And especially, I mean, it, it's been the conversation in FinOps since the beginning of FinOps, that getting this cost data in front of your engineers, in front of your engineering leadership, in front of your CFO in a way that is intuitive, in a way that is aligned with your organization, helps people [00:26:00] make better decisions.

Now you're applying that same concept to the data center, and you're applying that same concept not only to cost, but to carbon, and that really just ties a bow on the importance of getting this data and helping people make the right decisions on top of it

Mike Jaco: Yeah, absolutely. And I, I didn't hit on one point that you made, and that was the interest of the engineers. And it's, you know, funny you mention that because that was actually part of the business case in that, you know, yes, you know, we save carbon with, with costs, and so will the additional carbon, you know, uh, sustainability data help us save more money?

That was the question that leadership asked us. "Great, we, we want this data, and now, great, you found it and told us how much it costs to get it. But can that also help save more money than, than if we not?" And, and we argued yes, and that we had data and, and case studies, you know, that supported that. And I will tell you, after implementation, and I, you know, don't have the data that I [00:27:00] could show you, you know, kind of right now, but just in the conversations, we were meeting with, with programs, so engineering leads and, and, and engineering, um, uh, SMEs, you know, for FinOps, for, for their areas for months and months and months giving cost data.

And a lot of the conversations, not every one, but more than half of them shifted quite a bit when we started bringing sustainability data, too. That became more of the focus. Costs still were decreasing and savings, you know, were happened, but there was definitely an, an energy that, that, that, uh, the sustainability brought in a lot of situations that you could easily tell

Taylor Houck: I mean, that's the story right there. I mean, even if, let's say, the organization doesn't have the net zero goal that Mastercard does, it can still be a useful measure for driving engineering action. And I saw it firsthand at NBCUniversal, as I mentioned earlier, people that didn't care about cost, 'cause so much-- so many of the times we see in engineering organizations when you have the budget, and we're gonna get to AI very soon, where AI is blowing people's budget, so maybe it's not the case in [00:28:00] 2026.

But when you have the budget from last year to spend the money on cloud, what's the incentive for me to allocate engineering resources towards these cost optimization initiatives when I-- really it's, it's not a thorn in my side, we already have the budget, it's all good. But when you bring in the sustainability aspect and say, "Hey, you know, these idle machines that you've left running because you just don't, you know, don't have the time to allocate to deprovision them or, or verify that they really truly can be decommissioned, um, when I already had the budget to pay for them."

Well, what if you had a measure of the carbon impact of leaving them running, where people then feel more, let's say, personally responsible and, like, align with their, like, true sense of, like, self and identity that they want to do the right thing. I mean, you could recycle as much as you want in your personal life.

You can turn off the lights when you leave the room. It's not gonna make a dent compared to the cloud resources that someone might be [00:29:00] running,

Mike Jaco: Oh, so I, I got two things that you, you-- So you are a hundred percent right, and that is funny that you said that because we put a little slide together that, that showed that, that exact point that one engineer making this change is equivalent to, you know, fifty cars driving for a whole year, right?

You know, so we were able to then take that MT CO2 and put it in a real-world situation, you know, to say, "Hey, you just saved fifty cars worth of carbon driving for an entire year by doing this change."

Taylor Houck: It's amazing

Mike Jaco: one point, so, so absolutely, um, sustainability data energizes and br-- and changes the conversation. Um, when you talk about the cost budget, it's, it's... You, you said that, you know, we already have the budget. Why worry about this optimization? Take sustainability out and just focus on just cost. You have to have your buy-in, buy-in from finance, but our agreement with finance was the budget is there, so not going to take that budget. So then our argument to the [00:30:00] programs was, anything that you save, you get to reinvest.

Taylor Houck: Yep

Mike Jaco: know, right? So it's not, it's not that you have the budget, it's you can do more with the same budget. You can solve that efficiency, you know, right, that you haven't solved for yet. You can now increase the capacity for this that you haven't been approved for the budget of yet.

You can, you know, whatev-whatever that is. Um, and so that was motivating from that side.

Taylor Houck: And that's-- we did that a ton at NBC Universal as well. And what we, what we called it was self-funding growth.

Mike Jaco: Yep

Taylor Houck: Essentially, we would have, you know, this backlog of projects and initiatives that we didn't have budget for, and we would fund them through the savings of our cost optimization efforts in the cloud.

And actually, this is the perfect segue in getting to the hottest topic that everyone's talking about right now, which is AI. Because a lot of times what we're seeing at Point 5 is our customers are going out and they're saving a ton of money on their cloud footprint through identifying, you know, idle resources, through identifying misconfigurations and saving money, and then using that to fund [00:31:00] the AI growth that they're seeing.

Because right now in-- right, we're recording this in May of 2026, almost everyone that I talked to didn't plan for as much AI usage s- in the first half of this year as they're seeing come to life right now in front of them. Now, I want to pose two sides of the coin to you and get your perspective on one, the other or, or both, depending on where your interest lies.

But there's this idea of AI and managing AI spend, right? Then there's the idea of using AI to make the FinOps, the cloud optimization, the hybrid, you know, TBM FinOps, uh, persona more effective and more valuable. How do you see AI impacting this role of, of [00:32:00] managing the value of technology as we move forward?

Mike Jaco: I mean, talk about managing the value of technology. I-- It's only gonna increase, you know, it. And talk about the first side of it, which is kind of managing the expense, you know, right? And how do you, how do you do that? And the concept of tokens and stuff like that, right? Different from cost. You're just adding another complexity measure, you know, kind of into it to be able to compare an apples to apples, you know, type of thing. Um, but, you know, the same concept that, that, that we took from a kind of like a product level, we brought that, you know, to AI, and we had to try to define, well, what does that exactly mean? So where do we wanna separate? We-- The AI experts were the ones who, you know, we worked with. We didn't define like, "This is how we're gonna do it, and, and this is how, you know, we're gonna give it to you."

It was, "Okay, does this make sense to you? How do you want to see this? How do you want to run it?" And so we separated out like the data platform, right? And we were able to segregate that out. And that took work and tagging and stuff like that, but we wanted to separate that out, um, separately. So that'd be its own [00:33:00] product, you know, right? We want to separate out the SaaS products, Databricks, Snowflake, you know, right? And stuff like that. And, and really kind of carve that out as well and be able to allocate, you know, that stuff. Those are black boxes, you know. For most of them, you have to work with them to do it, but we found good partnership there. Um, and we were figuring it out together, you know, kind of how to do it. And then you have the, you know, your models that yourself that you're actually running. You know, right? So you have all three of those kind of components and pillars, and that's all around managing just the cost, you know, of it. And there was a, there was a great, um, win that we had to where we had anomaly alerts set up.

Um, and, uh, we caught, um, some spend, a hundred thousand dollar increase in two days and, and our alert caught it, and they stopped it immediately. Um, and even though-- So they had the budget for, I don't know, like a million dollars for like a whole month or something crazy like that, you know, right? Um, and so there wasn't, it wasn't anything flagging because they had the budget for it, but something had run off, you know, that they weren't expecting, you [00:34:00] know, right?

Now to

Taylor Houck: Right

Mike Jaco: if they would've, if they wouldn't, if we wouldn't have kind of caught this or whatever. And so that became a great, um, you know, use case and win and really put the importance on managing that cost and that expense. Then you're talking about then, okay, well, I think there's, there's maybe a, a third one.

So you're talking about how then FinOps can use it, but then you're also talking about how do you measure the value, the value that it's, the AI thing is bringing overall outside

Taylor Houck: 100%. Yep

Mike Jaco: you know, n-using it for FinOps, yes, you, you can, but that's only a small piece of it. What is all the other AI doing, and what is the value that is bringing?

I think FinOps can apply their same principles and same methodology to that stuff. We had not gotten that far, by the way, by any means at all. Um, but we were in that first part of just classifying kind of the expense. Um, in terms of using AI, then yeah, I mean, I think you, you start simple, right? It's, it's with the simple automation, the simple queries, the building of the dashboards.

You know, the, the, the, the simple analyzing of the data. Um, [00:35:00] I think the key thing that in AI in general, not just- With FinOps, that doesn't seem to be talked about enough is AI is only good as the person using it. You know, right? And you know enough, you and I, given the exact same AI with the exact same data and the exact same problem to solve, we will come up with different, you know, answers and different, whole strategies And everything, right?

Depending on how much detail you give it, how much, you know, do you make it personal? How much do you interact with it instead of is it a one-time prompt and in, or do you continue to build off of that? Are you able to sense that? That's not right. No, that's not what I meant. You know, instead of just taking it, right? AI is only as good as the user, you know, kind of using it, and there's something there that I think is kind of being lost, um, or not being focused on that I'm not sure how to quantify by any means. Um, but I think it's, it's careful when we talk about what AI can do and what it can replace, [00:36:00] then, you know, there needs to be, you know, kind of something, you know, there.

And you still have to train the next generation coming up

Taylor Houck: Yeah. And it's interesting because you mentioned, "Hey, I don't have all the experience in the world of, of doing this," but the reality is that no one does because it's happening, it's so new and it's happening so fast, right? And like one of the things that I've been thinking about is if you go back to the original problem statement that kind of created this FinOps domain, and you can call it FinOps, you can call it managing cloud spend, you can call it, you know, some people even look at it as a cloud architecture problem.

I, I tend to fall into that, that group, but really what it comes down to is that engineers were owning these purchase decisions that previously they relied on, you know, in the on-prem world. And, and part of this discussion today is that this is still the case, but when you have on-prem infrastructure, finance controls the purchasing of those servers, you have the LAN, you plug them in,

Mike Jaco: yeah,

Taylor Houck: all of that.

Mike Jaco: All that, yeah, taken care of for

Taylor Houck: You go to cloud and the engineers can spin up their servers. Well, with [00:37:00] AI, you're now seeing marketing departments using tokens. You're seeing, you know, IT is using tokens, HR is using tokens, operations is using tokens, and all these different people, it's almost-- you can almost equate it to hiring. And it's almost like if you just gave your, you know, all of your teams the ability to just hire freelance talent on demand to go do whatever work they want to do.

It's like,

Mike Jaco: Yep

Taylor Houck: we want that work to be done? What is the value of the work that is being done? How do we make sure that, you know, when we hire someone to do the work, we're not, let's say, hiring them to do the same repetitive task over and over again versus automating it being done? And what I mean by that is that I see a lot of people running, you know, the same general prompt or ideas through AI every single time burning tokens when they could have used AI to build an application that runs locally on their device that can do the same job, and then they only use the [00:38:00] tokens once.

Again, it's just an example, but measuring the value that AI is bringing is something that is going to be really important, especially as the spend is just increasing so rapidly.

Mike Jaco: And, and like you said, that's exactly what FinOps, what TB-- what, what it all boils down to is measuring the value. Visibility and measuring the value to increase the value, right? With the goal

Taylor Houck: Amazing. Amazing. Well, Mike, this has been amazing. Um, before I let you go, I do want to touch on a few more things with you because we've spent the majority or I guess all of the conversation up until this point learning from you. I now want to learn a little bit more about you and kind of change the topic from such tactical things to some more future-looking and, and perhaps you can even give some of the audience some, some advice or perspective.

But the first thing is, I understand that right now you are, um, between chapters professionally. I'm really keen to understand if someone with as much experience as you could go off and do so many different things, [00:39:00] and the industry is changing and, and is so dynamic right now. There's so many different avenues that you could apply your skill set and your expertise.

What's next for you, man? What are you, what are you looking for? What are you excited about right now?

Mike Jaco: Yeah. So I mean, honestly, the thing that excited to me about FinOps and doing FinOps right is transformational, right? And, and you can really transform how we do stuff, you know, the way it works, um, the value that we are bringing, and that's kind of mostly where I've, I've, I've kind of fallen, whether, you know, kind of be process transformation, organizational transformation, technology transformation.

It's just that transformation component is really-- I've realized how much I really like that. And ultimately, like that transformation is a people thing. The, the, the technology, the, the, the standards, the, the metrics, the data, those are all the tools, you know, right? But, you know, it's, it's really a people thing and, and that's what I enjoy. Um, so I think anything, you know, that any company or any group [00:40:00] that's looking to kind of transform the way they're, they're doing things, I'd be, I'd be interested to, uh, to look at.

Taylor Houck: Yeah. If anyone's listening and is looking for someone like that, you couldn't find someone better, um, than Mike. Now, I, I do wanna ask you one more thing just because I know you've had so much incredible experience in your career, and some of the people that are listening are earlier on in their journey.

When you look back at your career up until this point, what advice would you give to someone who's starting a FinOps role at some level of scale today?

Mike Jaco: So I think I'd, I'd start with, you know, kind of where I just said is that everything is a people problem. Um, you know, and you're trying to get an engineer to, um, you know, to do something, and they won't do it. You know, right? And you're like, "Why won't they make this change? It's so simple. It's, it's, it'll save so much money." You know, right? I would guess that ninety-nine point nine percent of chance is that engineer is not just like, "Hey, I want to waste money. I want to, to waste as much money as possible, and that's why I'm not doing it." Right? I [00:41:00] mean, that's, that's very unlikely, you know, to be the case. So why isn't that engineer?

You know, well, they're probably overloaded. They probably have different priorities. And so you're trying to get them to do something that might not benefit them, that might, you know, make their job harder. You know, like, you gotta understand that it's a people problem, right? You-- Even if you have to go talk to their boss to change the priorities, that's a people...

It becomes more of a people, you know, you know, relationship. And so Don't, don't get away from the people aspect, you know, of everything. Um, the other thing I would say, especially early in, your career, is you can't do everything. You know, right? You're, eager to prove yourself. You're eager to work and stuff like that, but you have to find the balance, and you have to find the trade-offs. and you know, I think that's where something that I learned is sometimes it's more important to talk about what you're not gonna do than it is what you're going to do, right? 'Cause prioritization is hard. Everybody wants everything, you know, right? And having that tough conversation and saying, "We can't have everything. Here's the trade-offs. Here's the balance," you know, [00:42:00] would be my biggest advice, I think.

Taylor Houck: Amazing. That is, that is such good advice. That is so pertinent, especially right now, Mike. Uh, if people are looking for more, where can they find you? Where's the best place to, uh, to reach out?

Mike Jaco: think the best place would be LinkedIn. Um, you know, message me, connect, um, and, uh, I'll, I'll definitely be responding

Taylor Houck: Awesome. Mike, this has been an incredible conversation. Thank you so much for coming on the show

Mike Jaco: Thanks, Taylor, for having me

Taylor Houck: And thank you to our audience. If you got something out of today's conversation, which I'm sure you did, please share this episode with someone else who needs to hear it. This has been another amazing episode of FinOps in Action.

We'll see you next time

Outro: That wraps up another episode of Fit Ops in Action. Thank you for joining. For show notes and more, please visit fit ops in action.com. This show is brought to you by 0.5, empowering teams to optimize cloud costs with deep detection remediation tools that actually drive [00:43:00] action.