Edge computing brings computation closer to users, devices and industrial systems. The figures below span EU policy targets and planning estimates, European cloud-flow models, global data-centre context, a narrow EU industry survey and peer-reviewed trials, so each number should be read with its stated population and measurement period.
Key Edge Computing Statistics
The most useful headline figures are:
- In the EU’s current Edge Observatory definition, an edge node provides latency below 20 milliseconds, a classification threshold rather than a measured deployment average.
- The EU Digital Decade target is at least 10,000 climate-neutral, highly secure edge nodes by 2030.
- Seven EU Member States were reported in 2024 as mobilising up to €1.2 billion in public funding for the IPCEI-CIS cloud-edge program.
- The same 2024 program was expected to unlock €1.4 billion in private investment in those seven EU Member States.
- EU Member State cloud data flows were estimated at 46,000 petabytes in 2024.
- Of that 2024 EU flow estimate, 36,600 petabytes were expected to move to cloud and edge capabilities within the EU.
- EU cloud data-flow economic value was estimated at €77 billion in 2024.
- EU cloud data-flow economic value is forecast to reach €328 billion by 2035.
- EU enterprises using cloud services generated approximately 30,000 petabytes per year of European cloud-based data flows in 2023.
- The EU roadmap estimated about 200 near-edge nodes would be required by 2025.
- The roadmap estimated around 6,000 far-edge nodes could be in place by 2025.
- The roadmap linked its planning scenario to 80% of data being processed at the edge by 2030.
- Global data-centre electricity consumption was estimated at about 415 TWh in 2024.
- Global data-centre electricity demand is projected to reach around 945 TWh by 2030 in the IEA base case.
- In a 2024 survey of businesses in the EU construction industrial ecosystem, 3% reported edge-computing adoption.
- A peer-reviewed edge energy-management deployment reported 145 milliseconds of mean system latency.
- The same deployment reported 81.2% lower bandwidth consumption than its conventional centralized comparison.
- A peer-reviewed federated-learning edge experiment measured 112.80 seconds for a mean FL round with two nodes in experiment 1.
Contents
- Definitions, deployment scale and EU targets
- Data flows and economic value
- Capacity, node types, power and investment
- Energy, emissions and wider data-centre context
- Adoption, outcomes and workforce signals
- Measured latency, bandwidth, energy and scaling
Edge Computing Definitions, Deployment Scale and EU Targets
The European Commission’s Edge Observatory uses a practical performance boundary: an edge node is compute capacity providing latency below 20 milliseconds. This is a definition for classification, not evidence that every deployed node achieves a particular average latency.
The same source sets a major public-policy ambition for the EU. By 2030, the Digital Decade target is at least 10,000 climate-neutral, highly secure edge nodes across the EU.
The investment picture also includes public and expected private capital. Seven EU Member States were reported in 2024 as mobilising up to €1.2 billion in public funding for the IPCEI-CIS cloud-edge program, with a further €1.4 billion in private investment expected.
Source: The Edge Observatory for the Digital Decade.
These are policy and program figures rather than a verified census of operating nodes or realized expenditure. The public figure is a mobilisation ceiling, while the private figure is expected leverage.
Edge Computing Data Flows and Economic Value
Edge statistics are often reported within a wider cloud-and-edge infrastructure picture. The European Commission estimated EU Member State cloud data flows at 46,000 petabytes in 2024, including 36,600 petabytes expected to flow to cloud and edge capabilities within the EU.
The destination estimate does not isolate edge-only traffic. It is therefore useful for understanding the scale of the digital infrastructure that edge systems may support, but it should not be read as an edge-network market size.
| Measure | Period and geography | Reported figure |
|---|---|---|
| Cloud data flows | 2024, EU Member States | 46,000 PB |
| Cloud and edge destination flows | 2024, EU | 36,600 PB |
| Cloud-flow economic value | 2024, EU | €77 billion |
| Cloud-flow economic value | 2035 forecast, EU | €328 billion |
For EU enterprises using cloud services, approximately 30,000 petabytes per year of European cloud-based data flows were estimated in 2023. That population excludes enterprises not using cloud services and measures cloud-based flows rather than all edge traffic.
Across Europe including the EU, EFTA countries and the UK, the 2024 European ECBDF economic value was estimated at €107 billion. Its components were €77 billion in the EU, €8 billion in EFTA countries and €22 billion in the UK.
Source: Data Flow and Economic Value EU Framework: Modelling Update and Data Collection.
The European Commission’s visualization tool covers 1,100 data points on cloud-flow volume, value and forecasts. That is a tool inventory count, not a count of edge nodes.
Edge Computing Capacity, Node Types, Power and Investment
The European Cloud/Edge Technology Investment Roadmap separates near-edge and far-edge planning assumptions. For 2025, it estimated that about 200 near-edge nodes would be required, with roughly 1 MW of IT power assigned to each node and about 200 MW of total near-edge IT power.
Far-edge planning used smaller and more distributed units. Far-edge nodes may need to sit less than 100 kilometres from customers; the roadmap estimated around 6,000 could already be in place by 2025, including about 4,000 that might need to be closer than 50 kilometres.
| Node or capacity measure | Roadmap period | Estimate or assumption |
|---|---|---|
| Near-edge nodes required | 2025, EU | About 200 |
| Near-edge IT power | 2025, EU | About 200 MW |
| Far-edge nodes | 2025, EU | Around 6,000 |
| Far-edge nodes under 50 km | 2025, EU | About 4,000 |
| Far-edge IT power | 2025, EU | About 350 MW |
| Total new cloud-edge power capacity | Through 2025, EU | 900 MW |
Far-edge nodes were modelled at 30–100 kW each, with the far-edge fleet potentially providing about 350 MW of IT power by 2025. The roadmap estimated 900 MW of total new power capacity through 2025, or more than 200 MW per year.
The same roadmap estimated that 80% of data could be processed at the edge by 2030. This is a scenario estimate, and the meaning of “at the edge” affects how the percentage should be interpreted.
For comparison, the roadmap put cloud data-centre capacity at about 350 MW by 2025, using 5–7 MW small facilities and 20–30 MW large facilities. It estimated €2.9 billion for that cloud infrastructure package and €2.1 billion for the near-edge infrastructure package.
Source: European Cloud/Edge Technology Investment Roadmap.
These roadmap values are planning estimates, forecasts or infrastructure-package estimates. They are not verified installed capacity, universal specifications or complete deployment costs.
Data-Centre Energy, Emissions and Edge Computing Context
The available energy statistics cover all global data centres, not edge-only facilities. In 2024, global data-centre electricity consumption was estimated at about 415 TWh, representing about 1.5% of global electricity consumption.
In the IEA base case, global data-centre electricity demand is projected to reach around 945 TWh by 2030. That scenario places data centres at just under 3% of global electricity consumption and projects approximately 15% annual growth from 2024 to 2030.
The IEA says data-centre electricity growth over 2024–2030 is more than four times faster than growth in total electricity use from other sectors. These figures describe the wider infrastructure context in which edge computing operates, not a standalone edge-energy total.
Global data centres produced around 180 Mt of indirect CO2 emissions from electricity use in 2024, excluding backup generation. They accounted for about 0.5% of global fuel-combustion emissions.
| Global data-centre context | Period | Figure |
|---|---|---|
| Electricity consumption | 2024 | About 415 TWh |
| Share of global electricity | 2024 | About 1.5% |
| Electricity demand | 2030 IEA base case | Around 945 TWh |
| Indirect electricity emissions | 2024 | Around 180 Mt CO2 |
| Share of global fuel-combustion emissions | 2024 | About 0.5% |
The IEA base case projects data-centre indirect emissions to grow almost 80% over the decade and rise to about 1% of global fuel-combustion emissions later in the decade. Backup-power emissions are excluded, and edge-only emissions are not separately identified.
Sources: Energy and AI: Energy demand from AI and Energy and AI: AI and climate change.
Edge Computing Adoption, Organizational Outcomes and Workforce Signals
A narrow sector survey offers one direct adoption signal. In 2024, 3% of businesses in the EU construction industrial ecosystem reported edge-computing adoption.
The same population showed broader adoption of related technologies, but those measures are not substitutes for edge adoption:
- Cloud adoption rose from 27% in 2023 to 37% in 2024 among businesses in the EU construction industrial ecosystem.
- IoT adoption rose from 12% in 2023 to 23% in 2024 in the same ecosystem.
- AI adoption rose from 10% in 2023 to 16% in 2024 in the same ecosystem.
- Big-data adoption was 14% in 2023 and 15% in 2024 in the same ecosystem.
- Augmented- and virtual-reality adoption was 11% in 2023 and 14% in 2024 in the same ecosystem.
- Robotics adoption rose from 5% in 2023 to 9% in 2024 in the same ecosystem.
Source: A blueprint for a competitive transformation of the construction ecosystem.
These are survey results for a defined construction population, not EU-wide enterprise adoption rates. The source table does not provide a 2023 edge-computing baseline, so the 2024 edge figure should not be converted into a year-over-year trend.
A separate IBM Institute for Business Value report identifies organizational findings about edge computing, but the supplied material does not provide additional quantified results from that report. It is therefore not used here as a numerical claim: Why organizations are betting on edge computing.
Measured Edge Computing Latency, Bandwidth, Energy and Scaling
Peer-reviewed trials show why edge results must be tied to a workload and setup. In a residential, commercial and industrial energy-management deployment, mean system latency was 145 milliseconds, bandwidth consumption was 81.2% lower than the conventional centralized comparison, and monitored energy use was 23.6% lower.
| Setting | Response time | Energy reduction | Peak-demand reduction | Operating-cost reduction |
|---|---|---|---|---|
| Residential | 138 ms | 18.5% | 21.3% | 23.6% |
| Commercial | 156 ms | 22.4% | 27.8% | 27.3% |
| Industrial | 182 ms | 16.8% | 19.5% | 24.1% |
The same study reported renewable-energy utilization of 32.7% in residential, 28.4% in commercial and 29.5% in industrial settings. Its study deployment period was not specified on the indexed page, and its results are not universal benchmarks.
A separate peer-reviewed federated-learning edge experiment measured how mean FL-round latency changed with node count:
| Experiment | 2 nodes | 4 nodes | 8 nodes |
|---|---|---|---|
| Experiment 1 mean FL-round latency | 112.80 s | 212.19 s | 675.52 s |
| Experiment 2 mean FL-round latency | 6,951.02 s | 9,258.29 s | 12,918.49 s |
These measurements come from a specific federated-learning implementation and experiments, not from a general edge-computing scaling law. The underlying paper is Enabling Federated Learning at the Edge through the IOTA Tangle.
For additional context, the supplied research also identifies a peer-reviewed simulation of edge-computing architectures in 5G networks, but it does not provide a separate quantified result used in this article: Energy-focused simulation of edge computing architectures in 5G networks.