Why market-size estimates disagree: a worked cloud-market comparison

Two estimates, one question

Two research firms published different estimates for what each called the global cloud infrastructure services market in the final quarter of 2025. Synergy Research Group put it at $119.1 billion. Omdia put it at US$110.9 billion.

The difference is about 7.4% of Omdia's estimate. That matters when a market-size figure informs an opportunity assessment, a competitive comparison or a forecast. It does not establish which estimate is better.

To investigate, we compared the original releases and supporting public documentation, reproduced the arithmetic and traced selected downstream references. We can identify differences in the published claims and test possible explanations. We cannot reconstruct either firm's full model or assign the entire gap to a specific cause.

The claims, side by side

Synergy's release appeared on 5 February 2026. Omdia's release followed on 26 March. Both describe worldwide spending or revenues for Q4 2025.

These are the firms' published estimates, not independently observed market totals. Our arithmetic treats Synergy's dollar symbol as US dollars; Omdia explicitly specifies US$. Neither inspected release provides the full currency-conversion method.

Compare definitions before comparing totals

Infrastructure as a service, or IaaS, provides computing resources such as virtual machines and storage. Platform as a service, or PaaS, supplies services used to build and run applications; see AWS's overview of these service models. Omdia also names bare-metal services, containers and serverless computing. Synergy explicitly names hosted private cloud.

Different labels may describe overlapping services. A container or serverless offering could sit within another firm's PaaS category. Naming these services separately does not establish that Omdia counts more. Naming hosted private cloud does not prove Synergy alone counts it.

A comparability ledger records what the evidence supports before an explanation is chosen.

Sources: the original releases; Synergy's current infrastructure and SaaS descriptions; Omdia's Cloud Ecosystems overview; and the historical Canalys definition below. Current descriptions provide context, not a frozen record of every Q4 2025 rule.

Private cloud remains an open question

A tempting explanation is that Synergy includes private cloud while Omdia excludes it. The inspected material does not establish that.

An original Canalys release from November 2022 included IaaS and PaaS on dedicated hosted private infrastructure as well as shared public infrastructure. It excluded direct SaaS spending while retaining the infrastructure used to operate those applications.

This is historical evidence from Canalys, now part of Omdia. It does not prove the same rules applied unchanged in 2025. It does provide a reason to investigate rather than assume private infrastructure was excluded.

Likewise, “managed private cloud” is a service-category label. It does not establish that all professional-services or consulting fees count toward the market total.

Reproduce the calculations

Let S = 119.1 and O = 110.9, in billions of dollars.

Using Synergy as the denominator gives approximately 6.9%. Both percentages describe the same dollar gap; the denominator needs to be named.

If each total was rounded to the nearest $0.1 billion, the underlying difference would remain between approximately $8.1 billion and $8.3 billion. This is a sensitivity calculation under an explicit rounding assumption, not a confidence interval.

If an additional category were the only difference between otherwise identical measurements, it would need to contribute roughly $8.2 billion in that quarter. We have not measured such a category. Assigning that amount to hosted private cloud, SaaS or neocloud revenue would turn a subtraction into an unsupported market finding.

Check the basis of a growth rate

Assuming Omdia's 29% growth and current total use the same basis, and treating the printed growth rate as exact:

This is our calculation, not a directly observed prior-year source value. Assuming conventional rounding of the total and growth rate widens the implied range to approximately $85.6–86.3 billion.

The same operation is not valid without further evidence for Synergy's 30%. That growth figure backs out major currency effects. 119.1 / 1.30 is therefore not a validated nominal Q4 2024 baseline.

Synergy's earlier Q4 2024 release printed $90.6 billion. Comparing that older figure with $119.1 billion produces about 31.5% growth in printed dollars. It does not contradict an FX-adjusted rate of 30% or prove a historical revision: the currency bases and release vintages first need to be aligned.

Use the vendor shares as a diagnostic

The releases also publish different vendor shares. Multiplying each share by its own market total gives an implied amount attributed to that vendor within the estimate.

Source shares: Synergy and Omdia.

Implied amount = published total × published share / 100. These amounts are our calculations, not company revenue disclosures. Matching the vendor labels does not establish that the firms assign the same products or revenue streams to them.

Suppose Omdia's vendor amounts remained unchanged and only the denominator increased to Synergy's total. Google's share would become:

Synergy instead prints 14%. Under nearest rounding of totals to $0.1 billion and shares to a whole percentage point, the implied amounts for Amazon and Google still do not overlap between releases. Microsoft's do. The calculation notes and notebook show the interval boundaries.

The test limits a possible explanation: adding a segment outside unchanged leading-vendor amounts cannot reconcile the full pattern. Vendor-level scope, allocation and measurement need examination too. The public evidence does not identify a complete reconciliation.

Amazon's own disclosure gives AWS Q4 segment sales of $35.579 billion. That is useful primary evidence, but neither analyst's market numerator is established as the entire company segment. A close numerical match cannot settle the definition.

A further denominator check matters within Synergy's release: its 68% top-three share concerns public IaaS and PaaS. Its full-market shares sum to approximately 63%. Those percentages describe different scopes.

What is confirmed—and what remains unresolved?

Different estimates can both be defensible when they measure different markets or apply reasonable methods differently. We have not established that these conditions fully explain this pair of figures, nor that either estimate is wrong.

For practical use, choose the source whose documented scope best matches the question, retain its date and limitations, and keep a time series consistent. Do not silently splice quarters from different firms or average their totals into a supposedly better estimate.

Trace each claim to its origin

Coverage of Synergy's figure traces back to its original release. Omdia's figure also appears in distributed press-release coverage. These repetitions help readers find the claims; they do not provide additional market measurements.

Following the source can catch changes in transmission. A TechTarget article gives Google 15% in prose and 14% in its chart; Synergy's original says 14%. We use the original figure and retain the inconsistency in the source record.

For the headline estimates in this comparison, the downstream pages lead to two statistic origins.

A checklist for comparing market-size estimates

1. 01

   Measure: Is the number revenue, customer spend, shipments, capacity or run rate?

2. 02

   Period and vintage: Which period does it describe? When was it published or revised?

3. 03

   Geography and customers: Which places and populations count?

4. 04

   Inclusions and exclusions: Which products and services count? Is an exclusion explicit?

5. 05

   Category mapping: Do differently named categories overlap? Do shared labels conceal different boundaries?

6. 06

   Vendor universe: Which companies and business lines count?

7. 07

   Accounting: How are supplier revenue and customer expenditure reconciled, including margins, taxes and gross/net treatment?

8. 08

   Currency: Which FX and inflation bases apply to the level and growth rate?

9. 09

   Method: Is the figure observed, modeled, annualized or forecast? What uncertainty is disclosed?

10. 10

    Lineage: Are apparently separate references repeating one underlying claim?

11. 11

    Calculation: Which transformations are valid? Have denominators, assumptions and rounding been stated?

12. 12

    Uncertainty: What remains unresolved, and what evidence would resolve it?

Use the comparison worksheet to record the answers. For this case, a genuine reconciliation would require matched vendor/product definitions, quarter-specific coverage lists, currency rules and a bridge between the totals. More pages repeating the releases would not supply that evidence.

Sources and methodology

This comparison uses public material checked on 30 September 2026. We inspected the original analyst releases, supporting program descriptions, a historical Canalys PDF and Amazon's disclosure; calculated differences and share diagnostics; and traced selected downstream references.

The linked figures are attributed source claims. Calculations and interpretation are New Eruditions' work. We did not access subscription datasets, conduct interviews or independently measure the cloud market.

The supporting resources are the source dossier, complete ledger, calculation notes, executable notebook and source-lineage map.

Limitations

The historical definition does not establish unchanged 2025 rules. Current program descriptions may have changed since the quarter studied. Product allocations, complete vendor lists, accounting bridges, currency policies and revision histories could not be fully aligned from the inspected public material. Rounding sensitivities are assumptions, not published confidence intervals. These limits prevent a complete numerical explanation of the $8.2 billion gap.

Read more about New Eruditions' approach to evidence, or use How to write a useful research brief to frame a question.

Corrections and updates

For a factual correction or broken source link, use our contact page or email private@neweruditions.com with the article address and the issue.

The original publication date remains fixed. A substantive change receives its actual update date and a visible note explaining a material correction. Formatting or routine rebuilds do not refresh the article's publication or last-updated date. The research-check date changes only after a substantive evidence review.