FinOps FOCUS specification becomes the common language for AI cost accountability

As artificial intelligence drives a new wave of token-based spend across cloud and on-premises environments, enterprises are discovering that understanding what they actually owe — and to whom — is harder than ever. The FinOps FOCUS specification is emerging as the open standard that gives practitioners the shared data language to cut through multi-provider billing chaos.

Fragmented cost and usage data across cloud providers has been the quiet tax on FinOps teams for years, forcing practitioners to build bespoke normalization pipelines before any analysis can begin. That pain is familiar ground for Karl Kraft (pictured), senior manager of software engineering at Walmart Inc., who has contributed to the FOCUS specification since its earliest pre-1.0 draft and now helps steer its continued development.

“When I joined the FinOps practice back before it was called FinOps, I was handed some Excel files of our cost and usage from our cloud service providers — two different cloud service providers,” Kraft said. “I looked at it and thought, what am I going to do with this? … I don’t want everybody to have to go through that pain again and again.”

Kraft spoke with theCUBE’s John Furrier and Paul Nashawaty at FinOps X 2026, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed how the FOCUS specification is lowering the barrier to multi-cloud cost allocation, accelerating chargeback at scale and preparing enterprises to account for AI token economics. (* Disclosure below.)

FinOps FOCUS specification unlocks chargeback and AI cost clarity

The FOCUS specification solves a deceptively simple problem: making cost data speak the same dialect regardless of which provider generated it. One practical example is how FOCUS handles amortized commitment-based pricing. Rather than surfacing a lump monthly charge that teams cannot allocate, the specification introduces an effective cost column that distributes a flat-rate subscription across daily consumption so individual teams see exactly what they used, Kraft explained.

“There is the effective cost — it’s a new column,” he said. “I can see over the month, my $30,000 is a thousand dollars a day and who consumes it each day, rather than just a big upfront cost that … I don’t know what to do with it.”

That same logic extends to Kubernetes workloads. The FOCUS 1.3 specification introduced provider-defined split cost allocation, enabling cloud providers to expose pod-level cost data natively inside the billing feed rather than requiring teams to instrument open source projects themselves, Kraft noted. With the dev cycle now open on version 1.5, the FOCUS specification working group is targeting a stock-keeping unit pricing dataset to standardize how cloud providers communicate pricing options — an area where even a single provider can surface three conflicting dataset variants today.

AI is accelerating both the urgency and the ambition of the roadmap. The FOCUS community has introduced contributor guidelines for AI agents within its own development workflow, using them to surface subtle inconsistencies across a specification too large for any single contributor to hold in memory. Looking ahead, version 1.5 will include features designed to break down AI spend by token type and workload, giving practitioners the granularity they need to tie model inference costs back to the teams consuming them.

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

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of FinOps X 2026:

(* Disclosure: TheCUBE is a paid media partner for FinOps X 2026. Sponsors of theCUBE’s event coverage do not have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE