Three insights you may have missed from theCUBE’s coverage of FinOps X

AI costs are becoming one of the most difficult aspects of enterprise AI adoption.

Unlike traditional cloud or software-as-a-service spend, AI costs are shaped by dynamic usage patterns, model behavior and external interactions, making it harder to keep investments aligned with business value. As enterprise AI adoption grows, organizations are reevaluating traditional cost governance models, according to Marco Meinardi, vice president analyst at Gartner Inc.

“With AI, now we’re dealing with spending sources that are even outside of our organization, and I’m not just talking about agents that have potentially endless loops of reasoning,” he said in an interview. “We’re also dealing with end users, our customers … and how they use our AI application — how they prompt them — is going to influence our costs. We’re dealing with two different problems that will require different solutions.”

Meinardi spoke with theCUBE’s John Furrier and Paul Nashawaty at FinOps X 2026, during an exclusive broadcast on theCUBE, SiliconANGLE’s livestreaming studio. FinOps leaders gathered at the event to discuss the growing challenges of AI cost governance and the need for new frameworks to measure, govern and manage AI costs as adoption accelerates. (* Disclosure below.)

Here’s the complete video interview with Marco Meinardi:

Here are three insights you may have missed from theCUBE’s coverage of FinOps X 2026:

Insight #1: FinOps is evolving beyond cloud cost management as token economics becomes the language of AI governance.

As enterprises expand AI deployments, the FinOps Foundation is extending its FOCUS specification to include AI spending. As AI usage grows across models, applications and infrastructure, standardizing cost data has become a prerequisite for governing AI costs, explained Matt Cowsert, principal product manager at FinOps, and Shawn Alpay, director of data engineering at the FinOps Foundation Project, a Series of LF Projects LLC.

“The data is not normalized across providers, across technology categories, public cloud, AI, SaaS, etc.,” Alpay told theCUBE. “Being able to tell that story with the same names of the columns, with the same definitions of the allowed values … being able to have that story told the same way across all providers — it’s incredibly valuable.”

The challenge extends beyond standardizing cost data. Token economics requires a fundamentally different operating model from the one FinOps developed for cloud spending, in the view of Nishant Gupta, chief availability officer at Salesforce Inc., and J.R. Storment, vice president and general manager of the Linux Foundation and executive director of the FinOps Foundation Project, a Series of LF Projects LLC.

“The key thing that’s different compared to traditional FinOps practice is just the completely different nature of how to think about tokens,” Gupta said in a keynote analysis. “It’s an abstract quantity. It’s very hard to relate tokens to a business outcome, very hard to relate input and output tokens in a predictable manner.”

As AI takes a larger share of enterprise technology budgets, FinOps leaders are shifting their focus from cost visibility to business value. That shift is highlighting the limits of simply applying existing processes to AI, pointed out Jennifer Hays, senior vice president and head of engineering excellence and technology strategy execution at Fidelity Investments, and Natalie Daley, director and global head of cloud economics and FinOps at HSBC Holdings PLC.

“There were a lot of enterprises and companies that just did a lift and shift [with cloud],” Hays told theCUBE in a keynote analysis. “They get the benefits of the cloud, but they do not necessarily take full advantage or the full value out of it. I think we’re going to see the same thing with AI.”

Here’s the complete video interview with Jennifer Hays and Natalie Daley:

Insight #2: Governance of AI costs is moving from visibility to autonomous control.

As AI costs become more difficult to predict and manage, organizations are increasingly turning to automation to identify anomalies, analyze spending patterns and connect costs to business outcomes. Amazon Web Services Inc. recently introduced a FinOps agent designed to monitor cloud costs, perform root-cause analysis and route alerts to the appropriate teams without waiting for end-of-month reporting, according to Jerry Rapisarda, director of AWS cost management and optimization at AWS.

“That’s really what the intelligence is about,” he said in an interview. “Understanding the context of your business and helping you to manage costs in the cloud.”

Governance controls are also becoming embedded directly into AI development platforms rather than operating as separate oversight processes. Microsoft Corp. is integrating model selection, content safety and security controls into developer workflows so organizations can apply governance and accountability as AI adoption scales, based on insights from Cyril Belikoff, vice president of commercial cloud and AI at Microsoft.

“We want to give developers the ability to innovate … but then have this data and AI platform that provides the guardrails that protect them from themselves,” he said during the event. “They can pick the right model, have content safety so that hallucination doesn’t happen, have security — and we expose all of those controls directly inside GitHub and GitHub Copilot.”

As AI costs spread beyond engineering teams and into broader business functions, organizations are looking for ways to apply policies without slowing innovation. Kion FinOps+ from Nor Labs Inc. is addressing that challenge through automated governance controls, emphasized Tatum Tummins, senior product manager at Kion.

“What it should mean is … if I’m an organization looking to implement this, I want to set a soft cap,” he told theCUBE. “When engineer Y hits a token threshold, I want to know about it, and then I have the decision — do I want to say, ‘Hey, keep going?’ Or do I want to actually talk about what we’ve done here? You just have to have some checks and balances.”

Google LLC has demonstrated how those governance principles can translate into measurable business outcomes. Through an internal initiative that applied orchestrating agents to supplier invoice reconciliation, the company increased throughput fourfold and generated $30 million in savings, noted Pravir Gupta (pictured), vice president and general manager of Google Cloud.

“That same pattern applies in so many different ways, because the trick here was not to roll out with a hundred percent accuracy,” he said in an interview. “The trick is that you have a human in the loop in the middle, where humans are reviewing the output of the agent and then providing the feedback.”

Here’s the complete video interview with Cyril Belikoff:

Insight #3: Sustainable AI investment requires a new operating foundation.

As organizations seek to fund growing AI initiatives without continually increasing spending, infrastructure modernization is emerging as a critical source of budget headroom. Aging hardware, low server utilization and inefficient architecture decisions consume resources that could be redirected toward new AI investments, explained Jim Greene, director of server product marketing at American Micro Devices Inc. and Mike Thompson, director of cloud product at AMD and governing board member of the FinOps Foundation Project.

“There’s a 30 to 40% [operating expense] difference between a couple of compute platforms that look the same,” Thompson told theCUBE. “A lot of folks don’t consider that nowadays, and particularly when you’re landing the applications, making those wise choices upfront is better.”

As AI costs expand across cloud and on-premises environments, organizations also look for consistent ways to allocate spending and establish accountability. The FOCUS specification is helping enterprises create a common framework for chargeback, cost allocation and AI spending transparency, noted Karl Kraft, senior manager of software engineering at Walmart Inc. and a longtime contributor to the FOCUS specification.

“There is FinOps for AI, and there’s AI for FinOps,” he said during the event. “FOCUS is a common nomenclature that’s going to really help the agents excel and process the data.”

The rapid pace of AI innovation is creating new challenges for budgeting, forecasting and long-term planning. Organizations must increasingly account for ongoing model changes in those planning processes, according to Trent Allgood, vice president of information technology asset management and FinOps consulting at SoftwareOne Inc. and governing board member at FinOps Foundation Inc., and Parker Nancollas, global FinOps practice lead at SoftwareOne Holding AG.

“Models need to be looked at as a consumable portion of what we’re building that’s replaceable and will be replaced,” Nancollas said in an interview. “Part of what you need to budget for and plan for now is … research and development for those new models.”

Here’s the complete video interview with Karl Kraft:

To watch more of theCUBE’s coverage of FinOps X 2026, here’s our complete video playlist:

(* Disclosure: TheCUBE is a paid media partner for FinOps X 2026. Neither the FinOps Foundation, the sponsor of theCUBE’s coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE