Back in 2015, TechViewz published The Internet of Many Types of Things, at a time when the Internet of Things still felt like a future made up of connected sensors, appliances, wearables, vehicles and industrial equipment.
That future arrived.
But IoT in 2026 is becoming something slightly different from what we imagined a decade ago.
The original idea was mostly about connectivity. Put sensors into physical objects, connect them to the internet, collect data and send that data somewhere useful.
Today, simply being connected is no longer enough.
The interesting shift is that many of those “things” are starting to process information and make decisions closer to where the data is created. Edge AI is pushing intelligence into cameras, industrial equipment, home devices, vehicles and sensors instead of sending everything back to a distant cloud first. That can reduce latency, improve privacy and allow devices to keep working even when connectivity is unreliable. (Synaptics)
In other words, IoT is slowly becoming AIoT: Artificial Intelligence of Things.
Even standards bodies are beginning to formalise that change. In 2026, the ITU published an AIoT reference model describing intelligence distributed across devices, edge systems and cloud platforms, including requirements for security, privacy, model integrity and resilience. (ITU)
The change is especially obvious in industrial environments.
Factories are no longer connecting machines only to monitor temperature, vibration or production output. Connected systems are increasingly being paired with AI for predictive maintenance, autonomous operations, robotics and digital twins. NIST’s 2026 roadmap for smart manufacturing highlights AI, advanced sensing, autonomous systems and digital twins as major parts of modern industrial environments. (NIST)
Smart homes are evolving in a similar direction.
The first generation of connected homes was basically remote control: open an app, turn something on, check a camera or adjust the thermostat.
The next generation is more about context.
A device may recognise who is in a room, understand environmental conditions, predict what is needed and take action without waiting for someone to tap a button. Technologies such as Matter are also trying to solve another old IoT problem: devices from different manufacturers not talking to each other properly. (Edge AI and Vision Alliance)
But making connected things smarter also creates a bigger security problem.
A basic sensor that reports temperature has limited influence. A connected device that can analyse its environment, communicate with other systems and make autonomous decisions has considerably more capability if compromised.
That is why modern IoT security is no longer just about changing default passwords or encrypting network traffic. Current guidance increasingly has to consider device identity, software integrity, secure updates, AI model integrity, privacy, resilience and what happens when devices begin making decisions themselves. NIST is still actively updating its IoT cybersecurity direction around these issues. (NIST)
Looking back at that old 2015 article, the prediction that the internet would connect many different types of things was correct.
What was harder to see was what would happen next.
The Internet of Things connected the physical world.
Now AI is teaching that connected world how to interpret what it sees.
The next phase may not really be about how many things are online.
It may be about how much authority we are willing to give them.