Skip to content
The Future Explorerby Benjamin Talin
Enablertechnological

Edge Computing

Doing computing close to where data is made, in a box at a factory, a phone mast or a local server, instead of sending everything to a distant data centre. That saves delay and data costs, which matters for machines that must react instantly. It is real in phone networks and factories but has not spread widely: only about 4% of computing work runs this way, and most firms still send data to the big building.

Maturity

3/ 5
measured
BasisGenuine production use in specific places - telecom networks, factory gateways, content caching - and it has not generalised. Edge and micro data centres hosted 4% of IT workloads in 2025. The Commission's own count of EU edge nodes was small and mostly test and research.SourceEurostat, Cloud computing statistics on the use by enterprises (data extracted Jan 2026); European Commission, State of the Digital Decade 2024 (Jul 2024); Uptime Institute, Global Data Center Survey 2025 (Jul 2025, INDUSTRY-BELIEF). The 75 per cent claim in the print text is a 2018 Gartner prediction - INDUSTRY-BELIEF with no primary basisVintage2024–2026
ReadinessCapability 4Deployment 3Absorption 3Bound bycapitalTrendflat

Where it feeds in

4 megatrends in 2 forces

Where you have already met this

In 2024 the European Commission put the number of edge nodes across the whole EU at an estimated 1,186, mostly used for testing and research. A year later, more than half of EU firms bought cloud services, and 85% of those used it for email. The data still travels to the big building.

If this moved

If running thousands of small sites got as cheap as running a few large ones, data could be processed inside a country that owns no cloud - a practical route to data residency. Digital sovereignty would still sit at 2, held by geopolitical alignment. The other three megatrends it feeds are held at 2 by regulatory clarity, managerial capability and workforce trust.

What to watch for

The share of IT workloads in edge or micro data centres in Uptime Institute's annual survey - 4% in 2025. Edge nodes in the Commission's count that serve paying customers rather than test beds.

Not to be confused with

Cloud computing. The cloud concentrates computing in a few very large sites; the edge spreads it over many small ones near the data. Most edge systems still report back to a cloud.

What it is

Processing performed close to where data is generated rather than in a distant datacentre.

How it works

  1. 01Machines produce dataSensors, cameras and industrial equipment send a constant stream over a local network.
  2. 02A nearby box catches itA gateway or small server on site, or inside a mobile network, receives the stream before it reaches the public internet.
  3. 03The decision is made locallySoftware on that box filters the data or runs a trained model on it, fast enough for a machine to act on the answer.
  4. 04Equipment acts without waitingThe result goes straight to a controller or actuator - no round trip to a distant data centre.
  5. 05Summaries go to the cloud laterFiltered results and logs are uploaded in batches for storage, reporting and retraining.

State of play

  • 52.7% of EU enterprises bought paid cloud computing services in 2025, 7.4 points more than in 2023 (Eurostat, 3 Feb 2026).
  • Of those buying cloud, 85% used it for email, 72% for office software and 72% for file storage (Eurostat, Statistics Explained, data extracted Jan 2026).
  • 84.7% of large EU enterprises bought cloud services in 2025, against 49.3% of small ones (Eurostat, data extracted Jan 2026).
  • The European Commission estimated 1,186 edge nodes in the EU, "predominantly utilised for testing and research" (European Commission, State of the Digital Decade 2024, Jul 2024).
  • Micro or edge data centres hosted 4% of IT workloads in 2025, against 48% in companies' own data centres (Uptime Institute, Global Data Center Survey 2025, Jul 2025 - INDUSTRY-BELIEF).
  • The Commission counts as an edge node any compute node answering in under 20 milliseconds (European Commission, Edge Observatory, updated 4 Aug 2026).

Why it matters

It makes latency-sensitive and bandwidth-heavy applications possible, and it changes where data physically sits - which is a jurisdiction question as much as an engineering one.

  • It is one of the few routes to data residency that does not require owning a cloud. That is why digital sovereignty leans on it - and why sovereignty is held at 2 by geopolitics, not by servers.
  • The measured deployment is a fraction of the talk: 4% of workloads, and a node count the Commission itself describes as mostly test and research.
  • It feeds autonomous mobility, precision everything and hyper-automation. None of them is waiting for it. Each sits at 2 on something human or regulatory.

What holds it back

Cost of many small sites

Every site needs power, cooling, security and someone to fix it. A central cloud spreads those costs over one building.

Fleet management

Updating and securing thousands of scattered machines is harder than running three large ones.

Size of the need

Most business computing is email, files and office software. None of it needs a 20-millisecond answer.

Terms

Edge node
A computer installed close to where data is produced, processing it there instead of in a central data centre.
Multi-access edge computing (MEC)
An ETSI-standardised design that places computing inside or next to a mobile operator's network.
OT gateway
A rugged box that connects factory equipment to an IP network and translates its protocols.
Edge inference
Running a trained machine-learning model on a local device rather than in the cloud.
Latency
The delay between sending a request and getting the answer, usually measured in milliseconds.

Steepens

The gradient behaves like a sticky rubber band: it is pulled hardest toward the weakest enabler. Marginal links are shown but do not pull.

If this moved one step

Computed from the map as it stands: this score raised from 3 to 4, every other score unchanged. It is arithmetic, not a forecast.

Raised one step on its own, it lifts no megatrend's weakest link; each gradient it feeds still rises a little.

The gradient also rises in

What if it moved?

What would change the gradient

Conditional, not predictive. No dates, no probabilities.

  1. 01Fleet management tooling that makes distributed operation tractable
  2. 02Data residency requirements, which are pulling deployment faster than latency is
  3. 03Hardware cost at the edge relative to centralised capacity

Common misconception

That it is a cost optimisation. Its more consequential property is that it lets processing happen inside a jurisdiction, which is one of the few practical routes to data residency that does not require owning a cloud.

Talin, B. (2026). Edge Computing. The Future Explorer, dataset v0.4.0 (2026-11-16). https://explorer.talin.digital/e/edge-computing/

Edge Computing: maturity 3/5 · The Future Explorer