Global data centre cost benchmarks

Benchmarking CapEx and OpEx in the global data centre market

As demand for cloud and AI infrastructure continues to accelerate, data centre developers, operators and investors are facing heightened scrutiny over cost discipline, location strategy and long‑term operational resilience.

While market demand remains strong, the economics of data centre development are becoming increasingly complex, influenced by construction market dynamics, labour availability, regulatory intensity, sustainability requirements and power constraints.

KPMG’s Data Centre Benchmarking Index: CapEx and OpEx in the Global Data Centre Market (2026) provides a structured, comparative view of how capital and operating costs vary across key global markets, and why those differences matter for strategic decision‑making.

Rather than focusing solely on headline build costs, the benchmark is designed as a decision‑support tool, helping stakeholders understand where data centre capacity can be developed and operated sustainably over a multi‑decade lifecycle.

Benchmarking CapEx and OpEx in the global data centre market

(PDF, 3.3MB)

Download report

Christopher Brown

Partner, Head of Strategy

KPMG in Ireland

mail
call

A consistent, comparable benchmarking framework

To support meaningful comparison across geographies, the benchmarking framework is based on a consistent set of assumptions. The analysis focuses on cloud‑only, greenfield developments using 2025 pricing inputs.

Bespoke AI, high‑performance computing and ultra‑high‑density deployments are intentionally excluded, allowing the benchmark to illustrate “steady‑state” market economics rather than highly customised or distressed project scenarios.

Inputs have been anonymised and aggregated to smooth above‑ and below‑market outliers and reflect representative cost profiles across developed economies. Abnormal items such as major grid reinforcements, exceptional planning obligations or unusual ground conditions are excluded, ensuring the results capture structural, rather than project‑specific, cost drivers.

While the current edition has a particular emphasis on European markets, the underlying cost dynamics explored in the index are globally relevant. Comparable patterns are observed across mature data centre hubs such as the United States, Singapore and Japan, as well as emerging growth markets in the Middle East and Asia‑Pacific.

Capital expenditure: construction drives regional divergence

Capital expenditure remains a critical factor in market selection, but the benchmark highlights that regional cost variance is driven primarily by construction dynamics rather than equipment pricing.

Across the markets analysed, total construction costs per megawatt vary materially. In the highest‑cost markets, construction pricing reflects a combination of elevated labour costs, constrained contractor capacity, complex regulatory requirements and deeper compliance obligations.

By contrast, lower‑cost markets typically benefit from broader subcontractor ecosystems, lower wage baselines and, in some cases, simpler planning or permitting regimes.

Owner‑furnished, contractor‑installed equipment (OFCI), including major mechanical and electrical plant, exhibits significantly less regional variation. Equipment pricing tends to be shaped by global original equipment manufacturer (OEM) supply chains, internationally traded components and increasingly standardised hyperscale specifications.

As a result, differences in mechanical and electrical equipment costs across markets are relatively narrow when compared with construction costs.

This distinction is important for investors and developers. While construction market conditions can shift materially between regions, equipment pricing is far more globalised.

As a consequence, headline CapEx comparisons should be interpreted with an understanding of what portion of cost variance is structural and what may be influenced by delivery strategy, contractor selection or timing.

Operating expenditure: labour is the primary structural driver

The benchmark demonstrates that operating expenditure varies even more significantly between markets than CapEx. For comparable facilities, annual OpEx can differ materially by geography, driven primarily by labour costs and staffing models.

Engineering full‑time equivalent (FTE) costs are the most visible contributor, but broader regional wage levels also affect security, cleaning and other soft‑services categories.

Many operational roles require continuous or 24/7 coverage, creating fixed staffing floors that do not scale linearly with megawatt capacity. As a result, per‑MW OpEx efficiencies are often limited beyond a certain site size.

While absolute operating costs are higher in some mature Tier‑1 markets, the benchmark also shows that OpEx expressed as a percentage of revenue tends to vary less.

Higher‑cost markets often benefit from stronger rental pricing, deeper demand and tighter capacity conditions, partially offsetting increased operating cost bases. This reinforces the importance of assessing operating economics in conjunction with revenue density rather than in isolation.

Other OpEx categories, such as facilities maintenance, IT and network operations, tend to show less regional variation due to global vendor pricing and standardised service models.

However, jurisdiction‑specific factors such as property taxation, insurance premiums and compliance requirements can still introduce meaningful cost differences over an asset’s lifecycle.

The AI effect and future capacity considerations

The benchmarking framework is intentionally based on cloud‑only specifications, but the index also highlights how emerging AI‑driven demand is reshaping data centre economics.

Higher rack densities and more compute‑intensive workloads are placing increased demands on power availability, cooling strategies, electrical resilience and operational expertise.

While AI‑optimised deployments are excluded from the core benchmarks, they are increasingly influencing site selection decisions even for conventional cloud facilities.

Power availability, time to connection and future upgrade potential are now assessed through an AI readiness lens, affecting both upfront design and long‑term operating models.

As these workloads scale, benchmarking approaches will need to evolve. Future editions of the index are expected to expand both the geographic scope and the range of deployment types assessed, reflecting the rapidly changing technical and economic landscape of the sector.

From build cost to lifecycle economics

One of the central conclusions of the benchmarking analysis is that data centre economics should be assessed through a lifecycle lens rather than focusing solely on initial development cost. The lowest‑cost market to build in is not always the most attractive market to own and operate over a 20‑year horizon.

Structural factors such as labour availability, power pricing, grid resilience, regulatory stability and sustainability obligations often have as much influence on long‑term returns as construction cost differentials. In capacity‑constrained or high‑demand markets, operators may also accept higher upfront or operating costs where revenue density and strategic positioning justify the premium.

For developers, operators and investors, this reinforces the importance of portfolio diversification, balancing exposure to premium Tier‑1 hubs with structurally lower‑cost markets that may offer stronger operating margins over time.

Supporting informed decision‑making

KPMG’s Data Centre Benchmarking Index is intended to support more informed, evidence‑based decision‑making across the sector. By isolating structural cost drivers and illustrating how CapEx and OpEx interact across regions, the benchmark provides a clearer foundation for assessing market attractiveness, investment strategy and long‑term competitiveness.

As the global data centre market continues to evolve, understanding these cost dynamics will be critical for organisations seeking to deploy capital efficiently, manage operational risk and align development strategies with future demand.

Get in touch

If you would like to discuss the findings of this benchmarking analysis, or explore what they mean for your data centre strategy, please get in touch with our team.

We would be happy to discuss your specific requirements and support your decision‑making.