Edge Computing: the architectural layer that makes the Smart Factory possible
Not a gateway, not a device: the point where production meets enterprise software.
When discussing Smart Factories, the spotlight always seems to be on the same technologies: Cloud Computing, Artificial Intelligence, the Internet of Things, and Big Data.
Edge Computing is often mentioned, yet rarely explained. Many people think of it as nothing more than a gateway that collects data from machines and forwards it to the cloud. In reality, it is much more than that.
Edge Computing represents one of the most important architectural layers of a modern Smart Factory. It is the point where the production environment (OT – Operational Technology) meets enterprise applications (IT – Information Technology), enabling machines, PLCs, SCADA systems, MES, ERP, cloud platforms, and Artificial Intelligence to work together efficiently.
To fully understand its value, imagine a very common manufacturing scenario. A production line includes Siemens PLCs, an industrial robot, a machine vision system for quality inspection, a SCADA platform for plant supervision, and a Manufacturing Execution System (MES) responsible for managing production orders. All of these systems need to communicate seamlessly with one another.
This is precisely where Edge Computing comes into play.
Edge Is Not a Device
One of the most common misconceptions is associating Edge Computing with a specific piece of hardware. In reality, Edge does not identify a product.
It can be deployed directly on a machine, on an Industrial PC, on a plant server, on an industrial gateway, or distributed across multiple Edge nodes located in different production areas.
What defines an Edge platform is not the hardware it runs on, but the role it performs: processing information as close as possible to where it is generated. This is why today we talk about Edge Computing rather than simply industrial gateways.
What Happens Inside an Edge Platform?
Imagine that the PLC signals the start of a new production run. That information is not simply sent to the cloud. A modern Edge platform can perform a series of processing operations locally.
It can read data through OPC UA or Modbus TCP. It can normalize it and convert it into a common format. It can publish it to an MQTT broker so that other services can use it. It can update a local database. It can trigger a workflow that prepares the MES for the new production run. It can send information to the SCADA system to update the process overview. It can synchronize data with Microsoft Fabric. It can even run an Artificial Intelligence model to determine whether the machine parameters are anomalous.
All of this happens within a few milliseconds. The Edge is therefore not simply a “bridge” to the cloud. It is a true integration and processing platform.
SCADA Is Not the Edge
Another common misconception concerns the relationship between SCADA and Edge Computing. Many people tend to regard them as the same thing. In reality, they serve different purposes.
SCADA systems are designed to supervise industrial plants: they display process diagrams, manage alarms, and allow operators to monitor production. The Edge, on the other hand, manages integration between systems. It can communicate with SCADA, but also with the MES, ERP, cloud platforms, and custom-developed applications.
In some architectures, the SCADA system can be hosted directly on the Edge platform. In many others, it remains an independent system that communicates with it. There is no single solution: there is only the architecture that best suits the specific production environment.
Where Does Artificial Intelligence Fit In?
The relationship between Edge Computing and Artificial Intelligence is another topic that is often misunderstood. Many people assume that AI must necessarily run in the cloud. In reality, the cloud and the Edge serve two distinct purposes.
The cloud is the ideal environment for collecting large volumes of historical data, building Data Lakes, training Machine Learning models, and analyzing production performance through platforms such as Microsoft Fabric. Once a model has been trained, however, it can be deployed to the Edge. This is where inference takes place.
The Edge uses the trained model to make decisions in real time. For example, it can detect anomalies, support quality control through Computer Vision, classify production events, or assist operators during maintenance activities.
The cloud trains.
The Edge executes.
What About Existing Production Facilities?
A Smart Factory is rarely built from scratch. Most manufacturers already operate PLCs, SCADA systems, machine tools, and enterprise software that have been installed and expanded over many years.
Edge Computing allows companies to maximize the value of these existing investments. Through standard communication protocols, APIs, and integration services, technologies developed in different eras can be connected without replacing existing equipment.
This ability to bridge different technological worlds is what makes a gradual and sustainable digital transformation possible.
Where Can an Edge System Be Deployed?
One of the questions we are asked most frequently is: "Where should an Edge system be installed?" The answer is simple: it depends on the architecture and the objectives of the project.
Contrary to what many people believe, there is no single "correct" location or standard device. Edge Computing is an architectural concept that can be implemented in different ways, selecting the solution that best fits the production environment.
On the Machine
When the machine manufacturer provides an Industrial PC or a dedicated computing platform, the Edge can run directly on the equipment itself. This solution is ideal for achieving the lowest possible latency or for integrating quality control, Computer Vision, and Artificial Intelligence directly into the production process.
On an Industrial PC or Gateway
One of the most common deployment scenarios. A dedicated device collects data from one or more PLCs, communicates with supervisory systems, and manages integration with the rest of the infrastructure. This approach is particularly effective for modernizing existing production lines without modifying the software running on the machines.
On a Plant Server
The Edge platform is installed on a local server that centralizes communications from multiple production lines, becoming the integration hub between OT and IT. It can host MQTT brokers, local databases, APIs, workflows, MES integrations, and AI services.
Distributed Architectures
In more complex industrial environments, multiple Edge systems operate together. Each production line has its own local Edge node, while a higher-level layer coordinates data collection, cloud synchronization, and centralized service management.
Ultimately, it is not the hardware that defines an Edge system, but the role it plays within the overall architecture. Its deployment always depends on the project's specific requirements, including response times, integration with production equipment, network availability, cybersecurity requirements, and the level of local processing required.
Why KyberEdge Was Created
Over the past few years—first at Rete Informatica and now at KyberEdge—we have designed numerous architectures for manufacturing companies. Throughout this experience, we learned that the real challenge is not collecting data. The real challenge is deciding where that data should be processed.
Some decisions must be made close to the production equipment within just a few milliseconds. Others require the computational power of the cloud, the analysis of massive amounts of data, and advanced Artificial Intelligence tools.
KyberEdge was born from this very experience. Our name reflects this design philosophy: bringing computing power and intelligence as close as possible to production processes while preserving the scalability and flexibility of the cloud.
For us, Edge Computing is not simply another technology to add to a project.It is the architectural layer that makes the Smart Factory possible.
