For over two decades, the conversation regarding IoT has been about connecting more devices. How do we get more things online? How do we enable communication between devices that, until recently, had no way to talk to each other? A large part of that question has now been answered.
Mohita Jaiswal - September 2019
Connected devices are everywhere. According to IoT Analytics, 21.1 billion of them are online today, projected to reach 39 billion by 2030. The standards have matured. Most modern homes can manage lights, locks, climate, and security from a single app. Factories run on connected production lines. Hospital systems share patient data across platforms that lived in silos just a few years ago. The frontier has truly moved.
The question now isn’t whether things can be interconnected — it’s what we do with the interconnectivity that already exists. The value sits in the layers above pure connectivity: how AI uses the data these devices generate, how telecom networks carry the volume, how the architecture stays secure, and how all of this reshapes what a product even means. That shift — from connectivity as the goal to connectivity as the foundation — is what we’re exploring today.
Why Interconnectivity Matters More Than Connectivity
Connectivity is when a device has an internet connection. Interconnectivity is when devices, systems, and platforms can exchange information meaningfully and act on it together. The difference sounds technical. It is, in fact, the difference between a smart device that’s useful and one that’s frustrating.
For instance, a doorbell with a camera is connected. A doorbell that recognizes the delivery person, unlocks the smart lock, sends a notification to the homeowner’s watch, and logs the delivery to a parcel tracking system is interconnected.
Most of these connected devices are in smart homes, factories, hospitals, vehicles, retail spaces, and city infrastructure. The connectivity layer is largely solved. The interconnectivity layer is where the real value, and the real difficulty, now sits.
How Do AI, Machine Learning, and the IoT Stack Fit Together?
For years, IoT meant data collection. Now it means what happens to that data the instant it’s collected. Artificial intelligence and machine learning are now the layer that sits on top of IoT, turning raw signals into automated action. The combined stack, sometimes called AIoT, uses pattern recognition to make sense of the firehose of real time data that connected devices generate. Let’s explore a few real-time applications:
1. Predictive maintenance
Manufacturing equipment fitted with vibration and temperature sensors feeds continuous data into ML models that detect anomalies long before a machine fails. The repair happens before the breakdown does.
2. Energy optimization
Smart buildings feed temperature, occupancy, and weather data into models that adjust heating, cooling, and lighting in real time. Hence, the energy consumption drops by 15–30% without anyone touching a thermostat.
3. Healthcare monitoring
Continuous-monitoring devices for chronic conditions feed clinician dashboards with ML-flagged anomalies — pattern shifts that would have been invisible in a once-a-month check-in.
4. Logistics routing
Fleet vehicles report location, traffic, weather, and fuel data to AI systems that rebalance routes in real time. Delivery times tighten without adding trucks.
How Telecom Networks Are Enabling the IoT Era
The original promise of IoT was held back for years by the telecom infrastructure underneath it. Setting up a new connected service used to require dedicated, single-purpose connections that took months to provision. That has changed.
Modern network connectivity for IoT now spans several distinct technologies, each suited to a different kind of device:
- 5G has become the standard for high-bandwidth, low-latency applications, such as connected vehicles, industrial robotics, AR/VR, and real-time video.
- 4G and LTE-M remain workhorses for general-purpose connected devices that need reliable but not ultra-low-latency communication.
- Wi-Fi and Bluetooth dominate inside homes, offices, and other defined spaces.
- Power wide area networks like NB-IoT and LoRaWAN serve the massive category of low-bandwidth, low-power devices, such as sensors, asset trackers, agricultural monitors, where battery life and range matter more than speed.
The bigger shift is in how networks are managed. eSIM and iSIM technology has removed the physical SIM card constraint for many IoT devices, letting them switch carriers and provisioning programmatically. The telecom industry’s role has evolved from selling connections to providing connectivity as a managed, programmable platform.
How Edge and Cloud Computing Process the Data
Some data needs to be acted on instantly. Other data needs to be analyzed deeply. No single architecture handles both well. This is the distinction between cloud and edge.
Cloud computing handles the heavy lifting in devices that require long-term storage, large-scale analysis, model training. When a manufacturer wants to find patterns across a year of operations data, the cloud is where that work happens.
Edge computing handles the time-sensitive work. It’s used for anything that can’t wait for a cloud round-trip. A self-driving car can’t send sensor data to a remote server, wait for a response, and then decide whether to brake. The decision has to happen at the device, in milliseconds.
The rule of thumb here is that one must process data close to where it’s generated when the cost of delay is high, and send it to the cloud when the value of deep analysis outweighs the latency. However, the newer wrinkle is edge AI. Small ML models now run locally on devices, processing data on-site without needing a constant connection to the cloud. For example, a smart camera that recognizes faces, a sensor that detects anomalies, a wearable that flags an irregular heartbeat — all increasingly do the work where the data is generated. The architecture isn’t cloud or edge anymore. It’s a layered system, with each component doing what it’s best at.
Where Interconnectivity Is Reshaping the Real World
The payoff of getting interconnectivity right is visible across several major domains:
1. Smart cities
Traffic systems, public transit, energy grids, and emergency services that share data in real time. A smart city that works isn’t the one with the most sensors; it’s the one whose sensors operate on a coherent data layer. Singapore’s smart traffic systems, Barcelona’s connected lighting and waste management, and a growing number of Indian smart city pilots all illustrate the point.
2. Smart homes
The promise made earlier was a thermostat that worked with a phone. The current reality, where the technology is mature, is whole-home energy management, security, and lifestyle automation, where the devices actually interoperate.
3. Manufacturing
Connected factories where machines, ERP systems, and supply chain partners exchange data in real time. The amount of data generated by a single connected factory in a day now rivals what an entire company produced in a year a decade ago.
4. Healthcare
Continuous monitoring devices feeding clinician dashboards and ML models that catch patterns no individual practitioner would notice. Wearables for cardiac patients, glucose monitors for diabetics, and continuous sleep trackers all reshape what ongoing care can look like.
5. Energy
Smart grids that balance supply and demand dynamically, with energy consumption data flowing from individual homes back to utilities for real-time pricing and load management. Renewable integration — solar, wind, EV charging — depends on this kind of grid-level interconnectivity.
So what is the pattern across all five? It’s that interconnectivity is what turns clever individual devices into a system that delivers value. Without it, you have a collection of expensive gadgets. With it, you have infrastructure.
The Security and Trust Layer
An interconnected world is a vulnerable one. Every device added to a network is a potential entry point. The same connectivity that makes IoT powerful makes it exposed.
The 2016 Mirai botnet, which weaponized hundreds of thousands of poorly-secured IoT devices to take down major internet services, was the first global warning. The 2021 Verkada breach, where attackers accessed live feeds from 150,000 security cameras across hospitals, schools, and factories, was the second. The pattern has only intensified since.
Securely connecting devices, networks, and data is now a design problem from day one, not a downstream IT concern. A few architectural principles have become baseline:
- Zero-trust networking: Every device authenticates every connection. No implicit trust because something is “inside” the network.
- Strong device identity: Each device has a unique cryptographic identity. Generic or shared credentials are no longer acceptable.
- Encryption in transit and at rest: End-to-end encryption for every transmission, regardless of distance.
- Continuous firmware updates: Devices that can’t be patched can’t be secured. Manufacturers who don’t support ongoing firmware updates are an unacceptable risk.
The NIST Cybersecurity for IoT Program provides foundational guidance for organizations designing and deploying IoT systems at scale. The principle worth carrying through is that security has to be designed into the IoT application from the architecture stage. Bolting it on later costs more, works less well, and leaves vulnerabilities the original design never accounted for.
What This Means for Business Models
IoT and interconnectivity haven’t just changed how products work. They’ve changed what counts as a product. A few new business models now made possible by interconnected systems:
1. Outcome-based pricing
Sell uptime, not equipment. Rolls-Royce sells jet engines by the hour of operation. Industrial equipment manufacturers increasingly sell performance guarantees rather than the equipment itself. The connected device reports its own performance — and that performance is what the customer pays for.
2. Hardware-as-a-service
Cars, appliances, agricultural equipment, and medical devices increasingly ship with software services that generate recurring revenue. The hardware is the entry point. The subscription is the business.
3. Data-as-a-service
The data generated by connected products becomes a product itself. Connected vehicle data, connected building data, connected supply chain data — each is a market unto itself, sold to insurers, planners, optimizers, and analysts.
4. Predictive service contracts
Instead of selling repairs after failure, sell preventive intervention before it. Maintenance becomes a continuous service rather than an emergency response.
