Hybrid-IT Is the Reality. Is FinOps Ready?
Eighty-nine per cent of surveyed FinOps practitioners operate in Hybrid-IT environments. The specifications, tools, and community conversations still centre on pure-cloud. Something has to give.
By Sima Niaztalkhouncheh · MSc Data Analytics, BSBI / UCA · Supervised by Dr. Anuj Batta
The FinOps discipline emerged from the cloud. Its foundational vocabulary, variable costs, continuous billing, tag-based allocation, showback and chargeback across accounts, reflects the operational realities of public-cloud infrastructure. Its specifications, most notably the FinOps Open Cost and Usage Specification (FOCUS), were developed to bridge the differences between cloud providers, not the differences between cloud and on-premises. And its community events, publications, and case studies overwhelmingly describe cloud-native organisations addressing cloud-native problems. Yet the empirical reality of enterprise infrastructure has, for years, been different. Most organisations do not operate in pure-cloud environments. They operate in Hybrid-IT, a mixed architecture in which on-premises systems persist alongside one or more cloud providers, each producing cost data under fundamentally different logics. A recent mixed-method study of twenty-eight FinOps practitioners quantifies the mismatch. Eighty-nine per cent operate in Hybrid-IT environments. The discipline they practise, in specification and community focus alike, was not built for the environment they actually inhabit.
The comparability problem
Cloud cost data are generated continuously. Every hour, every minute, every discrete resource event is captured by native billing APIs, tagged with resource identifiers, and made available for analytical consumption. On-premises cost data are generated episodically. A server purchase produces a capital expenditure event. That event is depreciated over an accounting schedule of three to seven years. Cost is imputed retrospectively, not measured continuously. There are no native APIs for on-premises cost. The data exists in asset management systems, capital planning spreadsheets, and manual reconciliation processes.
Comparing these two data streams within a single analytical frame is operationally difficult. Every practitioner surveyed acknowledged this. The mean Likert agreement that comparability is difficult stood at 3.66 out of 5.00 across the full sample, a majority-agreement result, and one that held across every demographic and role segment examined. The difficulty is not disputed by the practitioner community. It is not a niche concern of specialists. It is a defining condition of the environments in which FinOps is actually practised.
Two approaches to the same problem
The interview phase of the study surfaced two proposed approaches to bringing on-premises cost data into a FOCUS-compatible form. Each addresses the comparability problem from a different starting assumption.
The first, articulated by Erik Norman, a FOCP-certified CEO based in Belgium, treats the on-premises environment as fundamentally observable, provided the right instrumentation is in place. Erik’s proposal was concrete: hypervisor-level snapshots every five minutes, capturing VM utilisation at a frequency comparable to cloud metering. The output is a stream of usage data that resembles, in structure if not in origin, the continuous data cloud providers already emit. This stream can then be normalised to a FOCUS-compatible schema and analysed alongside cloud cost data.
“You take a snapshot of the VM every five minutes. You know CPU, memory, disk. You’ve got the same shape of data as the cloud APIs. Not the same volume, but the same shape.”
Erik Norman, FOCP-certified CEO
Rob Martin, VP of Member Strategy and Education at the FinOps Foundation, offered a complementary framing when asked about the on-premises extension. His observation drew a helpful distinction between the two hemispheres of infrastructure data. Cloud data represents the wire, the physical resource in use. Data centre data must represent both the asset (its capital value, depreciation schedule, and useful life) and the use (how the asset is consumed by workloads).
“You’ve got two sides of it. There’s the asset valuation, which is a finance question. And there’s the chargeback part, which is really where FOCUS for data centre lives, capturing the use messaging side of it.”
Rob Martin, VP Member Strategy & Education, FinOps Foundation
The FinOps Foundation’s FOCUS for Data Center working paper, referenced by Rob in the interview, formalises the second half of this framing. It positions FOCUS as an appropriate representation for the use-messaging side of on-premises cost, while explicitly not attempting to serve as a replacement for financial asset accounting. This is the pragmatic path. On-premises can be brought into FOCUS not as a full replica of cloud, but as a chargeback-oriented representation of use, layered onto separately maintained asset records.
The industrial complication
A third interview, with Engineer Ahmadi in the EPC (engineering, procurement and construction) sector, added a further complication that the cloud-native FinOps literature does not typically address. In capital-asset-intensive industries, oil and gas, heavy manufacturing, mining, telecommunications infrastructure, data centre capacity is frequently allocated at the project level rather than the workload level. A specific project acquires a proportion of the total data centre’s capacity, valued at the book-value cost of the underlying assets.
This model does not map cleanly onto either the cloud paradigm or the hypervisor-snapshot proposal. It is neither continuous-usage measurement nor asset depreciation in isolation. It is project-level allocation weighted by book value, and it is how a substantial portion of on-premises data centre cost is actually managed in industries that dominate global capital expenditure. The FOCUS specification, and the broader FinOps discipline, does not currently offer a first-class representation of this allocation model.
“In our industry, we allocate cost by project. Not by workload. Not by resource. By project. And the value we allocate reflects the book value of the assets that project consumes.”
Engineer Ahmadi, EPC sector practitioner
The implication is that Hybrid-IT is not a single problem. It is a family of related problems, each shaped by the industry, regulatory environment, and capital structure of the organisations that operate the underlying infrastructure. A FOCUS-aligned analysis that works for a technology-sector Hybrid-IT deployment may not work for an EPC-sector deployment without substantive extension. The specification is a foundation. It is not, by itself, a complete solution.
The semantic layer question
A further refinement emerged from the interview with Beau Nelford, an architect of the FOCUS specification. Beau observed that FOCUS standardises the structure of cost data, the columns, the field definitions, the taxonomies used to categorise resources and services. It does not, however, standardise the interpretation of derived metrics computed on top of that structure. Two teams starting from identical FOCUS-aligned data can compute different values for the same KPI, for example, effective savings rate, because the calculation rules themselves are not part of the specification.
Beau’s suggestion was that FOCUS may need to be paired with what he termed a semantic layer, a companion specification that defines how the standardised data should be interpreted for common analytical purposes. This semantic layer would sit above FOCUS, not within it, preserving the specification’s focus on data structure while adding a second layer of standardisation around derived interpretation.
“FOCUS gets us to a place where the data looks the same across providers. What it doesn’t get us to is a place where the analysis looks the same. That’s a different problem, and I think it’s the next one.”
Beau Nelford, FOCUS architect, FinOps Foundation
For Hybrid-IT practitioners, the semantic layer question is particularly acute. The variety of allocation models, cost structures, and analytical requirements across industries means that FOCUS-aligned data can be analysed in many defensibly correct ways, and yield materially different conclusions. Without a semantic layer, cross-organisational comparability of FinOps analytics remains limited even where the underlying data structure is standardised.
And now, AI
While the Hybrid-IT bridge is still being built, a new cost category has arrived: Artificial Intelligence workloads. Rob Martin, in the interview, offered an important observation about how AI cost data differs from traditional cloud cost data.
“AI data looks much more like operational telemetry than cloud billing data. Tokens are not really the thing you’re tracking, they’re more like packets in the old days. The future is layered data sets, not one giant matrix.”
Rob Martin, VP Member Strategy & Education, FinOps Foundation
The implication for FOCUS is structural. A specification designed to unify cloud billing across providers may not be the right shape for AI cost management, which requires higher-volume, telemetry-like data, often sampled rather than exhaustively captured. Rob’s expectation, that FOCUS will evolve toward a layered data model, with a core specification supplemented by domain-specific extensions, reflects the broader trajectory of the discipline: from a single-standard bridge across cloud providers to a family of coordinated standards addressing cloud, on-premises, AI, and whatever comes next.
For practitioners, the immediate question is which of these emerging strands to prioritise. The evidence from the study suggests a clear answer: start where the population is. Eighty-nine per cent of practitioners operate in Hybrid-IT environments. That is the population for whom the specification’s next-generation extensions matter most, and that is the population whose needs should shape the roadmap.
Where this leaves the discipline
The Hybrid-IT reality identified in this research does not require a wholesale reinvention of FinOps. It requires two coordinated moves.
First, the community conversation should recentre. Hybrid-IT is not a special case, a legacy concern, or a transitional challenge for organisations still migrating to cloud. It is the mainstream operating condition of the practitioner population. Content, case studies, working groups, and specification development should reflect this. When eighty-nine per cent of practitioners operate in a given environment, that environment is not the edge case.
Second, the specification roadmap should treat Hybrid-IT extension, semantic layer development, and AI tokenomics coverage as three coordinated priorities rather than three separate ones. Each addresses a limitation of the current specification. Each will fail as an isolated effort, because practitioners increasingly operate across all three domains simultaneously. The organisation running Hybrid-IT infrastructure today is also, in most cases, deploying AI workloads and grappling with cross-team analytical inconsistency. A specification that solves one of these problems in isolation solves only a fraction of the practitioner’s operational reality.
The FinOps discipline was founded on the recognition that cloud finance requires new thinking. Hybrid-IT finance requires the same recognition. The tools that emerged from the cloud era are foundational, but they are not sufficient. The practitioner community, the FinOps Foundation, and the vendor ecosystem now have both the empirical evidence and the specification substrate to move beyond the pure-cloud framing. The eighty-nine per cent are waiting.
About this research
This article summarises the Hybrid-IT strand of Sima Niaztalkhouncheh’s MSc dissertation, “Developing a Data-Driven FinOps Framework for Automated Cloud Cost Allocation and Variance Analysis in Hybrid-IT Environments using Machine Learning”, supervised by Dr. Anuj Batta and submitted at BSBI (Berlin School of Business and Innovation) in partnership with the University for the Creative Arts (UCA) in July 2026. The empirical work comprised a Machine Learning experiment on FOCUS-aligned Hybrid-IT data, an online survey of twenty-eight FinOps practitioners, and five semi-structured interviews with senior practitioners and standard-setters.
