The internet of things (IoT) and artificial intelligence (AI) hold great promise for the future, with many opportunities for innovation and affecting the way we do business. In this blog post I will explain why AI is needed closer to where the data are generated to make sense of a plethora of data coming our way, as well as why the Edge is the best place to handle this.


As AI finds its way to more and more devices, you need to consider a few things as it relates to your data, namely privacy and security, and closely aligned to this is data sovereignty. How much data do you want to liberate to the Cloud? If your data can be processed at the Edge, and made available per your requirements, do you need to send it to the Cloud?

Many IoT products currently use Cloud platforms for administering their services. Take the popular Amazon Alexa AI technology, for example. Amazon devices embedded with this AI interact with smart sensors in your house to automate lights, security systems, garage doors, and so on, ultimately leaking the pattern of your home occupancy to the Cloud. And, this valuable data are helping the hyperscalers build insight into your habits, which can be used to influence your purchasing decisions.

A better solution would be to have the same AI capabilities embedded in an on-premises IoT Gateway device to localize the data consumption, and thereby giving you control over your data. Eventually, with the right application, you can even monetize this data by selling it to third parties for building innovative applications.

Data Toll

Another advantage of localizing AI data processing at the Edge is that you keep unnecessary backhaul traffic to a minimum. Collectively, among the population, this can add up substantially. Furthermore, at an enterprise level, all of this can scale significantly, such that it may be prudent to consider an Edge solution for corporate applications. Per, there are about 3 million data centers in the United States, or 1 data center for every 100 people, and this number continues to grow. Hence, there is no shortage in the capacity to satisfy the growing demand of Edge Computing, with the added benefit of satisfying privacy and security requirements that the Cloud cannot satisfy.

Artificial Intelligence of Things

When “things” or devices like wearables, appliances, equipment, sensors, actuators, and other hardware products are connected to the internet, and subsequently interact with other devices to collect and process data at a machine-to-machine level, you essentially have the internet of things (IoT). Now, when systems learn and execute tasks in ways that can be perceived as smart, you basically get computed intelligence that is artificial. Hence when this type of intelligence is augmented to the internet of things, it means that those devices can systematically analyze and make sense of the available data, as well as act on that data programmatically. This is Artificial Intelligence of Things (AIoT).

Use Cases

Statistics on the growth in IoT space vary. According to International Data Corporation (IDC):

Worldwide IoT spending should reach over $1 trillion by 2022. While this is a big number, another report by McKinsey Global Institute (MGI) predicts that the potential economic impact of IoT in 2025 could reach between $3.9 to $11.1 trillion; these figures cut across many industry settings, with factories and cities accounting for about 50% of this figure. This translates to above 75 billion connected devices globally, and generating about 79.4 zettabytes of data by 2025 (source).

As of today, much of the data being generated is not consumed. For example, only 1% of data from an oil rig with 30,000 sensors is analyzed, and much of this is not even fully utilized; i.e., the data are mainly used for anomaly detection and control, not optimization and prediction, which provide the greatest value (source), and this is where AI can help.

Since applications using IoT devices have real-time constraints, and other concerns around data privacy and security as well as data sovereignty in certain settings, the Edge infrastructure is best suited for them. However, to make full use of the data generated by these devices, Edge Computing should be used to augment these applications with AI. Here are some practical examples of AIoT in market settings with the heaviest deployment of smart devices:

  • Factories and process plants can optimize their operations through predictive maintenance, inventory management, and adherence to health and safety protocols. AI can use historical data from sensors and actuators to build knowledge and experience on production outcomes, and suggest when to order replacement parts and when to perform services, as well as help regulate production while maintaining OSHA (Occupational Safety and Health Administration) standards.
  • Cities can improve public safety and health, resource management, traffic control, and so on. For example, intelligent transportation management systems can be used with real-time dynamic decisions on vehicle and pedestrian traffic flows. AI can analyze the data from cameras, lights, and other devices to make decisions about how best to alleviate traffic congestion with speed limit adjustments and/or traffic lights coordination.
  • Vehicles are starting to implement autonomous driving and navigation using, for example, autopilot systems that employ cameras, radars, GPS, and other sensors to gather data about driving conditions, as well as an AI system to make decisions from the information processed. Closely aligned with this is fleet management, where sensors and AI are used to better manage their assets through efficient fuel usage, vehicle maintenance, and driver behavior.
  • Retail environments have also started to implement conveniences like automated checkout using camera systems with AI that implement facial recognition technology. Intelligence about customers, like their preferences and mobility, can be analyzed to predict consumer behavior, and subsequently used to offer in-store personalized promotions, and even optimize store layout and operations. CRM and inventory management can also be enhanced through AI.

Together these industrial settings will account for about 73% of the aforementioned economic impact by 2025 based on the MGI report.


As the internet of things grows in this decade hyper connectivity will become the new norm, and along with it socio-economic changes will emerge and evolve. Just as Cloud Computing brought changes in how we conduct business, Edge Computing, alongside the technologies mentioned above, will introduce massive business model disruption. Analytical insights from AIoT among market settings will reveal trends and opportunities unlike what we’ve experienced in the past. The future is all about intelligent applications.