Cloud Migration Statistics 2026
By Axis Intelligence Research
Co-author: James Porter | Last updated: September 27, 2026 | License: CC BY 4.0
Axis Intelligence Research finds that only 14.86% of EU enterprises run their own software on purchased cloud computing power — even though 52.74% pay for cloud services. Across the EU, cloud-using firms are 3.02 times more likely to buy cloud email than cloud compute for their own applications. Most of Europe’s “cloud migration” is still SaaS, not workloads.
Quick Answer
Cloud adoption and cloud migration are not the same number. According to Eurostat’s 2025 ICT enterprise survey, 52.7% of EU enterprises paid for cloud services in 2025. But Axis Intelligence Research’s Own-Workload Reach (OWR™) — the share of all enterprises running their own software on bought cloud compute — sits at 14.86%. Denmark leads at 38.94%; Bulgaria trails at 4.01%.
Key Findings
- Axis Intelligence Research finds that 14.86% of EU enterprises ran their own software on paid cloud compute in 2025, versus 52.74% using any paid cloud service (Eurostat, 2025).
- According to Axis Intelligence Research, Denmark has the deepest workload migration in Europe, with an Own-Workload Reach (OWR™) of 38.94% of all enterprises in 2025.
- Axis Intelligence Research finds Italy ranks second in the EU for cloud adoption (75.60%) but fifteenth of 28 European countries for Own-Workload Reach (14.35%), the widest adoption-to-depth gap among large economies.
- According to Axis Intelligence Research, large EU enterprises run own workloads on cloud compute at 3.46 times the rate of small enterprises (OWR™ 43.42% vs 12.56%, Eurostat 2025 inputs).
- Flexera’s 2026 State of the Cloud Report (March 18, 2026) found 73% of organizations operate hybrid environments, confirming that full-estate migration remains the exception, not the default.
How Many Companies Have Migrated to the Cloud in 2026?
The headline number most pages quote — “half of European companies use the cloud” — is correct and almost useless for anyone planning a migration. It counts a 12-person accounting firm with Microsoft 365 the same as a manufacturer that moved its ERP, databases and custom applications onto IaaS.
Eurostat publishes both layers. We put them side by side.
| Metric (EU27, enterprises ≥10 employees) | Value | Year | Source |
|---|---|---|---|
| Enterprises using paid cloud services | 52.74% | 2025 | Eurostat |
| Enterprises using paid cloud services | 17.8% | 2014 | Eurostat |
| Change 2023 → 2025 | +7.4 pp | 2025 | Eurostat |
| Cloud users buying cloud e-mail | 85.15% | 2025 | Eurostat |
| Cloud users buying office software | 71.7% | 2025 | Eurostat |
| Cloud users buying file storage | 71.5% | 2025 | Eurostat |
| Cloud users buying security software | 65.5% | 2025 | Eurostat |
| Cloud users buying finance/accounting software | 58.2% | 2025 | Eurostat |
| Cloud users buying database hosting | 45.5% | 2025 | Eurostat |
| Cloud users buying compute to run their own software | 28.18% | 2025 | Eurostat dataset isoc_cicce_use |
| Cloud users buying a dev/test/deploy platform | 26.08% | 2025 | Eurostat |
| Own-Workload Reach (OWR™), share of all enterprises | 14.86% | 2025 | Axis Intelligence Research |
Adoption tripled in eleven years, from 17.8% in 2014 to 52.74% in 2025. The rise is real. What it measures is mostly the SaaS layer.
James Porter: Read the table bottom-up, the way a migration team would. E-mail and file shares are the lift everyone does first because the vendor does the lift for you. Compute for your own software is where the real work lives — the VMs, the database cut-over, the app that hard-codes a hostname from 2011. Only 28.18% of cloud-using firms have bought that. The rest bought an inbox.
What Is Own-Workload Reach (OWR™)?
Own-Workload Reach (OWR™) is an Axis Intelligence Research metric that measures the share of all enterprises in a market that run their own software on purchased cloud computing power. It separates workload migration from SaaS subscription.
Formula:
OWR™ = (share of enterprises using paid cloud services) × (share of cloud-using enterprises buying computing power to run their own software) ÷ 100
Worked example (EU27, 2025): 52.74 × 28.18 ÷ 100 = 14.86%.
Both inputs come from the same Eurostat survey wave (isoc_cicce_use, enterprises with 10+ employees), so the multiplication combines compatible denominators — Eurostat reports the compute figure as a percentage of cloud-using enterprises, which is exactly what makes the product valid.
A companion figure, the SaaS-to-Compute Gap, is the difference in percentage points between cloud users buying e-mail and those buying compute for their own software. For the EU it reads 56.97 pp (85.15 − 28.18).
Cite as: Axis Intelligence Research, Own-Workload Reach (OWR™), Cloud Migration Statistics 2026, axis-intelligence.com/cloud-migration-statistics/. License: CC BY 4.0.
Which Countries Have Migrated the Most Workloads to the Cloud?
This is the ranking nobody else publishes: 28 European countries sorted by how many enterprises actually run their own applications on cloud compute — not by how many pay for any cloud service.
Own-Workload Reach (OWR™) by country, 2025
| Rank | Country | Paid cloud adoption | Cloud users buying compute | OWR™ | Source |
|---|---|---|---|---|---|
| 1 | Denmark | 68.88% | 56.53% | 38.94% | Axis calc. from Eurostat |
| 2 | Lithuania | 58.06% | 52.79% | 30.65% | Axis calc. from Eurostat |
| 3 | Norway | 66.54% | 44.94% | 29.90% | Axis calc. from Eurostat |
| 4 | Sweden | 72.00% | 40.70% | 29.30% | Axis calc. from Eurostat |
| 5 | Malta | 75.07% | 38.79% | 29.12% | Axis calc. from Eurostat |
| 6 | Netherlands | 68.49% | 38.32% | 26.25% | Axis calc. from Eurostat |
| 7 | Belgium | 61.62% | 40.12% | 24.72% | Axis calc. from Eurostat |
| 8 | Estonia | 60.66% | 37.91% | 23.00% | Axis calc. from Eurostat |
| 9 | Finland | 79.21% | 27.78% | 22.00% | Axis calc. from Eurostat |
| 10 | Hungary | 47.96% | 36.73% | 17.62% | Axis calc. from Eurostat |
| 11 | Germany | 53.91% | 29.30% | 15.80% | Axis calc. from Eurostat |
| 12 | Portugal | 38.71% | 39.24% | 15.19% | Axis calc. from Eurostat |
| 13 | Austria | 52.08% | 28.95% | 15.08% | Axis calc. from Eurostat |
| 14 | Luxembourg | 47.70% | 31.57% | 15.06% | Axis calc. from Eurostat |
| 15 | Italy | 75.60% | 18.98% | 14.35% | Axis calc. from Eurostat |
| 16 | Ireland | 73.04% | 19.55% | 14.28% | Axis calc. from Eurostat |
| 17 | Spain | 41.65% | 32.85% | 13.68% | Axis calc. from Eurostat |
| 18 | Slovenia | 48.04% | 28.23% | 13.56% | Axis calc. from Eurostat |
| 19 | France | 40.41% | 33.48% | 13.53% | Axis calc. from Eurostat |
| 20 | Czechia | 54.90% | 24.11% | 13.24% | Axis calc. from Eurostat |
| 21 | Cyprus | 51.38% | 25.41% | 13.06% | Axis calc. from Eurostat |
| 22 | Latvia | 44.08% | 27.70% | 12.21% | Axis calc. from Eurostat |
| 23 | Croatia | 46.61% | 22.30% | 10.39% | Axis calc. from Eurostat |
| 24 | Slovakia | 36.44% | 26.80% | 9.77% | Axis calc. from Eurostat |
| 25 | Greece | 24.33% | 35.64% | 8.67% | Axis calc. from Eurostat |
| 26 | Romania | 24.94% | 30.14% | 7.52% | Axis calc. from Eurostat |
| 27 | Poland | 54.71% | 9.82% | 5.37% | Axis calc. from Eurostat |
| 28 | Bulgaria | 17.83% | 22.51% | 4.01% | Axis calc. from Eurostat |
| — | EU27 | 52.74% | 28.18% | 14.86% | Axis calc. from Eurostat |
Norway is included as an EEA country reported in the same Eurostat dataset.
Where adoption and depth diverge most
Three patterns stand out.
Finland has the highest paid-cloud adoption in the dataset (79.21%) and still ranks ninth on OWR™. Adoption is near-universal; workload depth is not.
Italy and Ireland are the sharpest cases of SaaS-led migration. Italy ranks second on adoption (75.60%, behind only Finland) and fifteenth of 28 on OWR™ (14.35%); only 18.98% of Italian cloud users buy compute for their own software. Ireland’s profile is almost identical: 73.04% adoption, 14.28% OWR™. Italy also posted the second-largest adoption jump between 2023 and 2025 (+14.2 pp, per Eurostat) — growth that the compute layer has not matched.
Poland runs the other way from what adoption suggests. At 54.71% adoption it sits above the EU average, but only 9.82% of Polish cloud users buy compute for their own software, the lowest figure in the dataset — an OWR™ of 5.37%.
At the top, Denmark and Lithuania are the only two countries where more than half of cloud-using enterprises buy compute for their own software (56.53% and 52.79%). Lithuania also recorded the largest adoption increase of any EU country from 2023 to 2025 (+19.7 pp, Eurostat), and that growth reached the workload layer.
James Porter: A high adoption rate with a low OWR™ tells you where the next migration wave is sitting: already-licensed tenants, already-trained users, and a server room nobody has touched. That is the cheapest migration pipeline in Europe on paper and the hardest one in practice, because the easy half is done and what’s left is the stuff with dependencies. If you sell migration services in Italy, Ireland or Poland, this table is your prospect list. If you run IT there, it’s your peer group — most of them haven’t moved the hard part either.
Do Large Companies Migrate More Workloads Than Small Businesses?
Yes — by a wide margin. Size is the single strongest predictor of workload depth in the Eurostat data.
| EU27 size class (2025) | Paid cloud adoption | Cloud users buying compute | OWR™ | Source |
|---|---|---|---|---|
| Large (250+ employees) | 84.67% | 51.28% | 43.42% | Axis calc. from Eurostat |
| Medium (50–249) | 66.78% | 34.05% | 22.74% | Axis calc. from Eurostat |
| Small (10–49) | 49.3% | 25.48% | 12.56% | Axis calc. from Eurostat |
According to Axis Intelligence Research, large EU enterprises run their own workloads on cloud compute at 3.46 times the rate of small enterprises. The adoption gap alone (84.67% vs 49.3%) understates this, because the compute layer compounds it: large cloud users are twice as likely as small ones to buy compute at all.
James Porter: Small firms don’t have fewer workloads to move. They have fewer people to move them. A 30-person company with one line-of-business app on a tower server under a desk is a textbook rehost candidate — and it’s also the company least likely to have anyone who has done a cut-over before. The small-business number is a staffing number wearing a technology label.
Why Do Companies Stay Hybrid Instead of Fully Migrating?
Outside Europe, no statistics office publishes the compute layer with Eurostat’s granularity, so the global picture comes from surveys. The most-cited is Flexera’s annual report, which in 2026 surveyed more than 750 cloud decision-makers.
| Metric | Value | Date | Source |
|---|---|---|---|
| Organizations operating hybrid environments | 73% | Mar 18, 2026 | Flexera 2026 State of the Cloud |
| Organizations citing cloud spend management as a challenge | 85% | Mar 18, 2026 | Flexera |
| Estimated share of cloud spend wasted | 29% | Mar 18, 2026 | Flexera |
| Organizations with a Cloud Center of Excellence | 71% | Mar 18, 2026 | Flexera |
| Organizations with a FinOps team | 63% | Mar 18, 2026 | Flexera |
| Large enterprises spending over $5M per month on cloud | 76% | Mar 18, 2026 | Flexera |
| Organizations running active workloads on AWS / Azure | 83% / 79% | Mar 18, 2026 | Flexera |
| Organizations using generative AI | 81% (72% in 2025) | Mar 18, 2026 | Flexera |
Flexera reports waste rising to 29% for the first time in five years, which it attributes to AI workloads. For migration planning, that matters: workloads moved without rightsizing carry their on-premises overprovisioning into a metered bill. Our FinOps statistics cover the cost-governance side in depth.
A methodology note that changes how these figures should be read: Flexera’s sample is cloud decision-makers at organizations already in the cloud, many of them large. Eurostat’s is a statistical sample of all enterprises with 10+ employees. Flexera’s 73% hybrid figure and Eurostat’s 14.86% OWR™ are not contradictory — they measure different populations — and Axis Intelligence Research deliberately does not merge them into a single number.
James Porter: Hybrid isn’t a halfway house anymore; for most estates it’s the destination. Mainframe batch, latency-bound plant systems, licences that cost more in someone else’s data center — those stay. The mistake is treating the 73% as a failure to finish. The real question is whether what stayed on-prem stayed on purpose.
Is Cloud Repatriation Replacing Cloud Migration in 2026?
The more accurate word for 2026 is re-routing. Workloads aren’t mainly coming back on-premises; a slice of them is moving between providers, driven by jurisdiction.
| Metric | Value | Date | Source |
|---|---|---|---|
| Worldwide sovereign cloud IaaS spending, 2026 forecast | $80 billion | Feb 9, 2026 | Gartner |
| Growth vs 2025 | 35.6% | Feb 9, 2026 | Gartner |
| Current workloads expected to shift from global to local providers | 20% | Feb 9, 2026 | Gartner |
| Sovereign IaaS spend from net-new solutions or legacy workloads awaiting migration | 80% | Feb 9, 2026 | Gartner |
Gartner’s estimate that 80% of sovereign IaaS spend will come from net-new builds or legacy workloads still waiting to be migrated is the single most useful migration signal published this year: it says the unmigrated estate is large enough to fund an entire market category. The Eurostat data shows where that estate sits. Our sovereign cloud statistics track the regulatory side.
Gartner also predicts (in its June 1, 2026 India cloud forecast) that by 2030 more than 60% of enterprises will run intensive AI model activity in one cloud while using their data in another — up from under 10% today. That same release forecasts India’s public cloud end-user spending at $17.5 billion in 2026, up 28.1% from $13.7 billion in 2025.
James Porter: “AI in one cloud, data in another” is a networking problem before it’s a strategy. Every one of those architectures pays egress and eats latency on the cross-cloud link. Migration planning that ignores where the data lands after the move is going to meet this problem in the invoice.
How Big Is the Cloud Migration Opportunity Left in Europe?
Axis Intelligence Research estimates the remaining headroom with one subtraction on the Eurostat inputs:
- Enterprises paying for cloud (52.74%) minus enterprises running own software on cloud compute (14.86%) = 37.88 percentage points of EU enterprises that already have a cloud contract but have not moved their own applications’ compute.
- A further 47.26% of enterprises (100 − 52.74) buy no paid cloud service at all.
This is an estimate of population headroom, not of spend or of workload count: Eurostat measures whether a firm buys a service, not how much of its estate runs there. It is still the clearest public indicator that workload migration in Europe is closer to its middle than its end.
For the provider side of the same market — who is capturing the spend — see our cloud market share statistics (quarterly revenue from filings) and cloud computing statistics hub. Sector-level budgets are in cloud spending by industry.
Cloud Migration FAQ
What percentage of workloads are in the public cloud in 2026?
No official statistics body measures the share of workloads in the public cloud. The closest public measure is the share of enterprises running their own software on cloud compute: 14.86% of EU enterprises in 2025, per Axis Intelligence Research’s OWR™ built on Eurostat data. Vendor surveys that quote workload percentages cover self-selected, mostly large respondents.
Which EU country is furthest along in cloud workload migration?
Denmark. According to Axis Intelligence Research, 38.94% of Danish enterprises ran their own software on paid cloud compute in 2025 (OWR™), followed by Lithuania (30.65%) and Norway (29.90%), based on Eurostat’s 2025 survey.
Is using Microsoft 365 or Google Workspace the same as migrating to the cloud?
Not in a workload sense. SaaS e-mail and office suites move the application, not your infrastructure. In Eurostat’s 2025 data, 85.15% of EU cloud users bought cloud e-mail but only 28.18% bought compute to run their own software — a 56.97-point gap.
Why do countries with high cloud adoption score low on workload depth?
Because adoption surveys count any paid cloud service. Italy (75.60% adoption, 14.35% OWR™) and Ireland (73.04%, 14.28%) show high SaaS uptake with limited IaaS use for firms’ own applications. Rehosting custom software requires skills, dependency mapping and cut-over windows that SaaS does not.
Are small businesses behind on cloud migration?
On workloads, yes. Small EU enterprises (10–49 employees) have an OWR™ of 12.56% versus 43.42% for enterprises with 250+ staff — a 3.46× difference, per Axis Intelligence Research using Eurostat 2025 inputs.
What share of companies run a hybrid cloud in 2026?
73% of organizations operate hybrid environments, according to Flexera’s 2026 State of the Cloud Report (published March 18, 2026, survey of more than 750 cloud decision-makers).
Will sovereign cloud requirements trigger a new wave of migrations?
Gartner forecasts sovereign cloud IaaS spending of $80 billion in 2026 (+35.6%) and expects 20% of current workloads to shift from global to local providers. It attributes 80% of that spend to net-new solutions or legacy workloads still awaiting migration.
How much cloud spend is wasted after migration?
Flexera’s 2026 survey estimates 29% of cloud spend is wasted, the first increase in five years, which Flexera links to AI workloads. Overprovisioned servers rehosted without rightsizing are a common contributor.
Methodology
Collection. Axis Intelligence Research retrieved Eurostat dataset isoc_cicce_use (Cloud computing services by size class of enterprise) via the Eurostat dissemination API on September 26, 2026, for reference year 2025 (survey conducted in the first months of 2025; data update February 27, 2026). Population: enterprises with 10+ persons employed, NACE sections C–J, L–N and group 95.1, excluding the financial sector, as defined by Eurostat. Flexera and Gartner figures were taken from the publishers’ own press releases, fetched the same day.
Indicators used. E_CC (enterprises using paid cloud, % of enterprises), E_CC_PCPU (computing power to run the enterprise’s own software, % of cloud-using enterprises), E_CC_PEM (e-mail, % of cloud-using enterprises), E_CC_PDEV (development/deployment platform).
OWR™ formula. OWR™ = E_CC × E_CC_PCPU ÷ 100. SaaS-to-Compute Gap = E_CC_PEM − E_CC_PCPU. Size-class multiples use the unrounded products. All arithmetic was run twice — in floating point and in exact rational arithmetic (Python fractions.Fraction) — with identical results.
Scope. OWR™ measures whether an enterprise buys cloud compute for its own software, not the volume of workloads it runs there. Countries’ 2025 figures for the Netherlands carry a Eurostat break-in-series flag on one sub-indicator (development platforms), which is not an OWR™ input. The metric is comparable across all EU/EEA countries in the dataset; non-European comparisons would require a survey with an equivalent compute indicator.
What we did not publish. We did not combine Flexera’s survey percentages with Eurostat’s statistical-sample percentages; the populations differ. We did not quote vendor “percentage of workloads in the cloud” estimates, which rest on self-selected samples.
About This Dataset
File: cloud-migration-statistics.csv — 274 rows, one observation per row, long format. Every row carries source_org, source_document, source_url, retrieved_date, is_primary, axis_calculated and, for Axis rows, a method_note with the formula.
Coverage: EU27 + 27 EU member states + Norway (2025); EU size classes (2025); global survey and forecast indicators (2026).
License: CC BY 4.0 — free to reuse with attribution.
Cite This Page
APA: Axis Intelligence Research, & Porter, J. (2026, September 26). Cloud migration statistics 2026: Europe moved its inbox, not its workloads. Axis Intelligence. https://axis-intelligence.com/cloud-migration-statistics/
MLA: Axis Intelligence Research, and James Porter. “Cloud Migration Statistics 2026: Europe Moved Its Inbox, Not Its Workloads.” Axis Intelligence, 26 Sept. 2026, axis-intelligence.com/cloud-migration-statistics/.
Chicago: Axis Intelligence Research, and James Porter. “Cloud Migration Statistics 2026: Europe Moved Its Inbox, Not Its Workloads.” Axis Intelligence, September 26, 2026. https://axis-intelligence.com/cloud-migration-statistics/.