The path to substantial deployment of IoT on industrial applications continues to be slow and full of roadblocks. Its most significant adoption so far has been as a replacement of the previous GSM Machine-to-Machine (M2M) connections, mostly carried out with cellular technology. M2M is still being used in many industrial applications that require the reliability and security of GSM, a technology that has been around for over 35 years.

Evidently, the data and management requirements that existed 35 years ago are not the same today. The number of connected machines has grown exponentially and so has the number of sensors and actuators installed in those machines.

Furthermore, a new wave of connected devices, fueled by the reduced cost and size of communications hardware and storage, and the data-hungry applications of today require a new way to communicate with those devices. Technologies such as LTE-M, Zigbee, NB-IoT and HaLow solve the connectivity problems. The upcoming 5G networks will also offer additional connectivity benefits for billions of devices.

According to Juniper Research, we are now reaching Phase 3 of industrial IoT adoption. Phase 1 was the initial M2M solution. Phase 2, when the first dedicated IoT connections and devices appeared, marked the appearance of small-scale projects, mostly replacing M2M as proof-of-concept.

Phase 3, which started just two years ago, is when IoT platforms started to emerge as an aggregation point for different data types and sources with support for application layers. At this stage, industries started to pay attention to the full potential of IoT and new business models and use cases began to appear. Many business applications are now rolled out and new projects start every day, although few of them have reached a scale of 100,000 units.

Phase 4 will be short—less than two years—as we are starting to see it in many industries. It begins when businesses recognize IoT as not just a technology but a new business transformation tool that requires rethinking industrial processes to execute. This phase involves all layers and departments of any industry to collaborate in a new strategic and technical alignment.

By 2020, Phase 4 will give way to Phase 5: As enterprises establish their vision for IoT and define new business models and requirements accordingly, innovative IoT projects will start to grow in size, with deployments of over 100,000 connected units across the entire organization. Phase 5 is when traditional industries will begin the transformation to full digital businesses.

By 2023, when Phase 5 of IoT transformation is fully adopted, industrial IoT will have over 46 billion active industrial connections. “This growth, equivalent to 140% over the next four years, will be driven by edge-computing services—the processing of data away from the cloud and closer to the source—increasing both deployment scalability and security,” says Juniper Research.

Obviously, Phase 5 will not be the end of the transformation of the enterprise by IoT. Phase 6, which will take a few more years—although we are starting to see the first glimpses of it—is when artificial intelligence (AI) will kick in as the real game-changer in industrial evolution. It will mark the end of IoT as mere devices being connected to the internet. New, powerful edge-computing devices and real-time computing and analytics will make parts of IoT fully autonomous and able to modify processes and systems without external control.

Juniper Research author Steffen Sorrell says, “IoT at the edge dramatically increases project scope and value. However, it must be noted that work around standardization, interoperability and how to manage the decentralization of data processing remains in development.”