FinOps breaks out of the cloud

FinOps finally became ubiquitous across the enterprise landscape last year with 75% of Forbes’ Global 2000 companies now all-in, according to IDC. And while the maturity of those practices varies, large organizations at the forefront of FinOps are scaling up and out, driving the cloud optimization practice into new areas of IT, including as a way to get a handle on spiraling AI costs.

Jeff Wysocki, CIO at mining firm Mosaic, recently hired his first full-time FinOps professional. “We’re 80 to 85% in the cloud and for us, the job is proactively tracking this spend, then educating developers and data teams on how to use cloud capabilities in a cost-effective manner,” he says.

By contrast, in some large financial services firms, the role of FinOps has broadened. “We’re moving FinOps practices into larger scopes of vendors,” says Jennifer Hays, SVP and head of engineering excellence and accessibility at Fidelity Investments. “We’re looking more at SaaS and PaaS, and our enterprise technology organization is trying to utilize some of the same principles for our on-premises data centers.”

And J.R. Storment, executive director of nonprofit FinOps Foundation, which develops best practices, education, and certification programs, adds that FinOps is about accelerating delivery, avoiding waste, and making more informed decisions on the trade-offs related to tech value.

Plus, many enterprises were doing FinOps well before the term was coined. Tigran Khrimian, chief technology engineering officer at the Financial Industry Regulatory Authority (FINRA), says he started developing best practices in 2014. The nonprofit regulates broker dealers and operates trade reporting systems that track vast numbers of transactions. “We never really adopted the term ‘FinOps,’ but we’ve been doing it from the beginning,” he says. And with 70-plus AWS services, 150,000 compute instances, and an exabyte of data, there’s a lot to manage.

“We averaged 612 billion events a day in 2024 and have had daily peaks as high as 900 billion,” he adds. “When you process big data, it gets really expensive really fast, so we had to form a team right away. We’ve built processes and automations for DevOps, operations, and dashboarding to determine what services have been underutilized and which ones had to be turned off.”

And for Shamim Mohammad, EVP and CITO at used car retailer CarMax, FinOps helps provide spend oversight. “Now we have a more granular understanding of our spending,” he says. “It provides transparency and drives accountability.”

FinOps practices evolving beyond the basics

For many organizations, FinOps is still about the basics of cloud cost control. When he was chief innovation and information officer at ICANN, Ash Rangan says his organization worked with a single cloud service provider (CSP), which made cost tracking easier. Today, as CEO of SaaS provider DoubleCheck, he feels the push and pull between what his customers pay, and forecasting how that’ll affect the company’s CSP charges on the back end. It’s all up to his head of product to figure that out. “I wouldn’t say we have complete control, but we can see what’s being provisioned, and when and what levers there are we can toggle,” he says.

For larger enterprises, it’s a whole different ballgame. In 2025 the scope of FinOps has expanded well beyond CSPs already. Financial services, technology, telecom, and retail firms are leading the way, according to the FinOps Foundation.

The Foundation is also responding to requests from practitioners to extend its framework to cover other areas of IT, such as private cloud, SaaS, licensing, and on-prem data centers. “We’ve had to make some quick changes to the FinOps framework,” Storment says. “FinOps is starting to become this reporting backbone across variable technology spends.” That interest is one reason why the FinOps Foundation community doubled to 60,000 participants last year.

CarMax, in addition, built a cross-functional team, reporting to the CIO, that uses many of the principles put forth by the FinOps Foundation. But his organization developed its own model, says Mohammad. While it’s not formally following the framework, “as it matures, we’re looking at it as a benchmark,” he says. And its FinOps principles aren’t limited to public cloud or SaaS. “Most FinOps principles have been applied to technology financials as a whole,” he adds.

Why the interest in extending FinOps? “A big spender on AWS might have billions of tiny charges in a single month,” Storment says. “After managing that volume of data for the cloud, it’s easy to add other types of spend.” That’s one reason why, in the FinOps Foundation’s 2025 State of FinOps survey of 861 of its members, 65% said their FinOps practices are also being asked to optimize spend for SaaS, as well as licensing (49%), private cloud (39%), data centers (36%), and AI (63%). In addition, according to the report, 97% of respondents are investing in multiple infrastructure areas for AI, which reinforces and accelerates how managing an organization’s AI spend will require taking on new areas of spend for FinOps practitioners.

“In response to changes in practice we’ve seen in mature organizations, we’ve added scopes so you can, if needed, apply those same principles and capabilities from managing cloud to SaaS and the data center,” Storment adds.

So to manage costs at scale, organizations must be able to react quickly, and FinOps practices should enable spend optimization decisions on a daily basis, rather than every quarter. “You’re spending money every second, so you need to be able to make these decisions quickly,” Hays says.

Applying FinOps lessons to contain AI costs

The rapid adoption and growth of AI in the enterprise has brought significant change and challenges. “Over the past two years, one of our areas of focus has been controlling costs in the AI/ML space,” says Hays. “We saw a large bump in usage and we’ve been working with the teams to understand their consumption and manage that, but I’m not sure we have enough tools that provide the right level of insight to effectively manage consumption in that space yet.”

Storment adds that this year is all about how practitioners are already thinking of how FinOps can optimize AI spend. The governance on AI costs is moving from ‘spend whatever you need,’ to ‘are we spending too much, and what value are we getting,’ he says.

So FinOps practices focus on both FinOps for AI — the optimization of AI usage — and AI for FinOps — leveraging AI to advance FinOps practices. Efforts to apply FinOps to AI are still in the early stages since the ways in which CSPs charge for AI continues to evolve. FinOps practitioners are also using AI to train on large data sets of billing and other data to identify opportunities for greater efficiencies, and can help with such things as tracking daily progress against monthly recurring charge commitments.

Many tools, many gaps

As FinOps efforts continue to scale, tooling has become a major focus, with 34% of FinOps Foundation survey respondents citing investment and tooling as a top practitioner need to achieve their priorities. Only organizational alignment ranked higher. And among the tool sets available, a few trends stand out.

FinOps tool vendors are focused on adding new AI capabilities, FinOps teams need the proper tools to analyze and optimize costs, and vendors provide a wide range of offerings in many different capability segments. IDC tracks more than 90 different tools in the FinOps space in its annual Market Glance of FinOps, says analyst Jevin Jensen. The report cites Broadcom’s Tanzu CloudHealth, IBM’s Cloudability (formerly Apptio), and ServiceNow Cloud Cost Management as FinOps platform market leaders based on the number of active customers and scope of offerings.

“They all provide multi-cloud cost optimization and resource optimization recommendations, and they all have different strengths and weaknesses,” Jensen says. “ServiceNow is tightly coupled to its Now Platform workflow engine and ITSM/CMDB products. CloudHealth has strong ties to virtualization and IT automation found in VMWare Cloud Foundation. And Cloudability ties to Apptio’s TBM products and IT financial management suites, including forecasting.”

Out of those 90 tools, however, just a few can handle IaaS, containers, and SaaS optimizations. “So if companies need all three, they’re likely to need more than one tool,” Jensen adds.

Hays also says you need a front end that helps you contain costs, and while CSPs have gotten much better at providing consistent billing data in accordance with The FinOps Foundation’s specification framework (FOCUS) — something FinOps tools can use — SaaS is another matter.

“SaaS providers aren’t part of the FOCUS movement as much, and every vendor we bring has a different consumption model and presents their billing information in a very different way, with different levers to help contain costs,” Hays says. Fortunately, she adds, they’re moving in the right direction. Meanwhile, Storment says the FinOps Foundation expects more SaaS vendors to adopt the FOCUS billing specification in the coming year.

But while FINRA is a big user of SaaS, the services are all priced in a way that’s highly inconsistent, which makes managing across them, according to Khrimian, a challenge. And even with such an abundance of tools, needs sometimes outstrip capabilities. For example, FINRA’s AWS services have different pricing methodologies for storage, compute, serverless, and other services, plus there are available discounts for things like committing to specific spending levels. “We couldn’t find a tool to manage that, so we wrote code to integrate AWS billing with our financial systems,” Khrimian says, adding that the team continues to build all of its reporting dashboards and forecasting tools.

Supply and increasing demand

In Q4 2024, 92% of enterprises said AI was an important capability they wanted in FinOps tools, IDC found, so many software vendors are adding it to their roadmaps — something Jensen says to look for in enterprise forecasts. And a few vendors have early releases. With so many tool choices, however, Jensen warns against tool sprawl since one third of enterprises are already using two or more FinOps tools, he says. 

For organizations with one CSP or those at the beginning of their FinOps journey, however, the CSP-provided tools may be enough for now. At DoubleCheck, the CSP-provided tools are sufficient to ensure observability and visibility, Rangan says.

Hays describes another challenge: “Platform-oriented costs like Kubernetes’ containerized platforms, and some things that support machine learning and AI on the major CSP platforms, are less mature than those for compute storage, databases, and those sorts of things.”

And don’t expect tool vendors to step up in terms of helping to maximize the value of your FinOps investment, warns Jensen. “Many cloud cost vendors don’t offer professional services to aid in maturing your FinOps people or processes,” he says, so you’re on your own if you want more than basic tool implementation.

Teams scale up as practices mature

FinOps teams have been expanding as their mandate has grown. Last year, the median team size increased from nearly six to eight people, according to IDC, and very large organizations may have 50 or more people. FinOps teams are often dominated by people with engineering backgrounds and use a hub-and-spoke model to disseminate best practices, tooling, standards, and recommendations to engineering teams.

“Total cost of ownership is done by the individual teams,” says FINRA’s Khrimian. “Our operation in the center is there to develop best practices and tools to assist delivery teams. We built competencies in the center with an eye toward enabling individual teams to adopt and own them.”

Khrimian has just under a half dozen people on his FinOps team. They have strong engineering credentials, but they’re also financially savvy. “We also have a parallel finance team that closely works with us,” he adds.

Fidelity also has a hub-and-spoke structure, where each business unit works with the Enterprise FinOps team. While engineers comprise most of her team, Hays also has a small number who are finance-focused. They review transparency and budgeting, and engage with the finance organization monthly to review trends. “We have a platform engineering organization and we help them set priorities,” she says. “We also have individual teams and squads — there are a few hundred of them.” The Enterprise FinOps staff handles contract negotiations, some compute purchases, and purchase agreements, while the business units focus on consumption management and optimization.

CarMax’s strategy has also evolved. It built a recommendation engine into its developer portal that shows usage trends, calls out cost anomalies, and makes daily recommendations. But one challenge for this year, Mohammad says, is to ensure the organization follows up on those recommendations, and that compute costs are part of the calculation when evaluating new initiatives. “This requires us to also educate product managers on an ongoing basis to keep costs top of mind,” he says.

​​Hays’ FinOps team has collaborated with core technology products and services that engineers across the business unit use to build software quickly and manage costs more effectively. “Any time you have tools that are reused, you have an opportunity to embed better behavior when it comes to optimization,” she says. “Those optimizations now apply to 80% of our business units technology, leaving our FinOps team to focus on the remaining 20%. It’s a win for platform engineering because they can tell how effective they are.”

FinOps and engineering partnership is key

FinOps is about running the technology organization like any other business, Mohammad says. As cloud and SaaS become an ever larger part of overall IT spend, “near real-time insights and shared accountability become more important than how we traditionally used to manage tech investments.”

The early wins with FinOps, when teams focus on consolidating contracts and purchase strategies, are the easy money, practitioners say. Savings that come in subsequent years take more work for possibly lower returns, so teams must move beyond the basics and work closely with engineering teams to analyze costs, and offer up more sophisticated and targeted tools that optimize cost efficiencies. “This year we’ve focused on Kubernetes optimizations and our analytical platforms where we have a large amount of spend,” Hays says.

Recently, her team built a resource shredder that automatically shreds and redeploys resources for containers where utilization isn’t high enough to warrant keeping them. Ultimately, she says, to optimize effectively, you have to think about how to do something on behalf of the engineers. “Our focus is on integrating our optimization activities in the products being created by the platform engineering organization,” she says.