The Second Half of Industrial Digitalization: From "Equipment Connectivity" to "Sensing, Networking, Computing, and Control"

Have you noticed a change?
In the past, the first thing a factory did for digitalisation was connect its equipment.
PLCs were connected, robots were connected, sensors were connected, industrial cameras were connected.
And then everyone felt: the digital foundation is built.But now, it may be exactly the opposite.

Equipment connectivity is only the first half of industrial digitalisation.

Why?
Because in the past, equipment connectivity solved one problem:
"Can the data be transmitted out?"
Now the real problem on the industrial site has become:

"After the data is transmitted out, who computes it? After computing, who decides? After deciding, who executes?"

This is why one change is increasingly worth attention:
industrial digitalisation is moving from "connected equipment" to "acquisition, network, computing, control."

Look at a production line today – it is no longer simply a PLC plus sensors.
There are industrial cameras, robots, AGVs, various sensors, and edge computing devices, while above them are connected MES, WMS, ERP, and even AI models and agents.

In the past, the industrial network only needed to ensure:
stable communication between equipment.
But today that is not enough.
Because once AI enters the site, what the network transmits is not just a few control commands and status data, but possibly images, time-series data, model inference results, and real-time interaction between agents.

More importantly, the location of computing is changing.
In the past, large amounts of data were sent to the cloud.
Now more and more judgments are being placed next to the equipment.
For example, when an industrial camera captures an image, it does not necessarily have to be sent to the cloud for judgment.
Inference can be completed directly on an edge industrial PC, and the result then handed to a PLC or robot for execution.

MIIT's "AI + Information and Communication" Innovation Development Implementation Opinionsreleased this year has explicitly proposed exploring cloud-network-edge-device collaborative inference to reduce inference latency and terminal computing requirements; the Industrial Internet and AI Integration Action Plan released earlier this year further proposed promoting the evolution of industrial networks toward "integrated control, network, and computing."

What does this show?
The network is no longer just "building roads."
It is beginning to become infrastructure connecting data, computing power, AI, and control systems.

01The Underlying Architecture of Future Factories

So the underlying architecture of future factories will increasingly look like this:

Equipment handles perception, the network handles connectivity, the edge handles computing, AI handles judgment, and control systems handle execution.

And one step further:
AI will not just judge – it will also participate in decision-making.

At this point, the value of the industrial network changes completely.

In the past we asked:
"Is this production line connected to the network?"
In the future, enterprises should truly ask:

"Have the data, computing power, and control capabilities of this production line been connected into a closed loop?"

Because only when they are connected does industrial AI truly have the opportunity to enter the production site.

02What Truly Matters About "Acquisition, Network, Computing, Control"

So what truly matters about "acquisition, network, computing, control" is not that another new concept has appeared.
Rather, it represents a very big change:

The foundation of industrial digitalisation is upgrading from "connecting equipment" to "connecting the entire production decision-making closed loop."

And this may be what truly needs to be reconstructed as the industrial internet enters the AI era.