Hornbill Smart Manufacturing"Industry 4.0" has been talked about for ten years, and most factories still hear it as "what does that have to do with me?" The reason is that it is usually presented as a vision rather than a checklist. In fact it has a very clear technical boundary: nine technology pillars, set out systematically by the Boston Consulting Group in its 2015 report Industry 4.0: The Future of Productivity and Growth in Manufacturing Industries, and carried into textbooks since then with the list barely changing.
Laid side by side, the nine are not about the future. They describe one chain along which data travels from the shop floor to a decision and back to the shop floor: first collect the data (connection), then turn it into judgment (decision), then turn judgment into action (execution), with a layer of security underneath holding it all up.
Connection layer: getting the data (3 pillars)
1. Industrial IoT (IIoT). Fit equipment, molds, material bins and AGVs with sensors and gateways so they can report their own state. This is the starting point of everything; without automatic collection, the other eight are castles built on manual data entry. The practical form is often modest: a current transformer on an old machine tool tells you whether it is really working. The trap is connecting only new machines, when the ones most worth watching are the old ones without communication interfaces.
2. Cloud computing. Move compute and storage from the server room to the cloud and scale on demand. The value is not that it is cheaper, but that several plants can look at data on one common definition. A hybrid setup is more common in practice: real-time control stays on site so it works in real time and survives a network outage, while analytics and reporting go to the cloud. The trap is putting real-time control in the cloud; when the network blips, the line stops.
3. Horizontal and vertical integration. Vertical means connecting device layer → SCADA → MES → ERP. Horizontal means connecting supplier → plant → customer. This is the least glamorous of the nine and the one that decides success. The real cause of death in most failed "digital projects" is not the algorithm; it is MES and ERP codes that do not match, two BOMs, and work order numbers that the systems do not recognize in each other.
Decision layer: turning data into judgment (2 pillars)
4. Big data and analytics. Move from after-the-fact reports to in-process alerts and advance prediction. Typical cases are predictive maintenance, correlation of process parameters with yield, and energy anomaly detection. The precondition is data quality. Data with mismatched stock records, scrambled timestamps and broken batch traceability will only give a model's elegant wrong answers.
5. Simulation and digital twin. Try it in the virtual world first, then do it for real. The levels differ greatly. Line layout and logistics simulation are low-cost and quick to pay back, and are the level most factories can really use. A real-time synchronized plant-wide digital twin costs a great deal and suits validation scenarios. The trap is trying to build a whole-plant twin at the start and ending up with an expensive 3D animation.
Execution layer: turning judgment into action (3 pillars)
6. Robots. From fixed-program arms to AGV/AMR and collaborative robots that sense their surroundings and work alongside people. The real barrier is usually not the robot, but the line layout, standardized bins and dispatching system around it.
7. Additive manufacturing (3D printing). The most solid value today is not mass production but tooling and fixtures, emergency spare parts and prototype parts: turning a two-week outsourced job into two days. Replacing batch injection molding or machining with it still does not add up on cost in most cases.
8. Augmented reality (AR). Remote expert guidance, assembly instructions and visualized maintenance. Of the nine, it is the easiest to make a demo work and the easiest to leave at the demo stage. A simple test: will front-line workers wear it every day?
The foundation: cybersecurity (pillar 9)
9. Cybersecurity. The better the first eight go, the more critical this one becomes: connected equipment, data in the cloud and an integrated supply chain all widen the attack surface. The factory is special because OT and IT have opposite logic. IT puts confidentiality first; OT puts availability and safety first, and one false-positive automatic isolation can stop a whole line. So it is not a matter of IT adding a firewall. It takes network segmentation, an asset register, remote access control and an emergency plan.
How to use the nine
Do not treat it as a shopping list; treat it as a self-assessment. For each pillar, ask three questions: what level are we at now, do we need to move next, and can we calculate the money saved if we do?
The order of adoption is fairly fixed. First 3 (integration) and 1 (collection) to connect the data, then 4 (analytics) and 5 (simulation) to put the data to use, and last 6, 7 and 8 on the execution side to automate actions. Security (9) runs alongside everything and cannot be added afterwards. The most common sign of the wrong order: a pile of robots and AR glasses bought, while data is still passed around in Excel.
You do not need all nine. Most mid-sized factories need to do three things properly: 1, 3 and 4. Choose the rest by industry. A simple way to judge whether a digital proposal is credible is to ask the vendor which of the nine they intend to touch and why they are leaving the others alone. If they cannot say, they are probably selling software, not solving a problem.
Translated from the original Chinese article on the WeChat account “犀牛鸟智造”.
