AI and IoT: a symbiotic relationship critical to meet enterprise needs

In this article, I consider the ways in which the latest technology hot topic, AI, and the last, the Internet of Things, have lots in common and are inextricably linked in terms of supporting enterprise transformation. Numerous applications – which is what enterprises really care about – will rely on both, and organisations focused on either will in many cases need to understand the other.

Over the last few years it has, of course, been impossible to miss the rise and rise of Artificial Intelligence as a business tool. Generative AI has certainly seized the crown of the Internet of Things as the hottest new technology area. We have also seen the increasing use of the term AIoT as a portmanteau acronym combining the two (although I have to say I'm not a big fan of it as a term). With all of that in mind, I think it’s worth reflecting on the ways in which those two technology areas are inextricably linked.

Adopters don’t really buy IoT or AI

One of the biggest issues with the Internet of Things was that it was really just a helpful shorthand term used by insiders to refer to a set of technologies used to remotely connect and manage devices and the data they generate. Adopters, however, never thought about it in those terms. Few of them thought of themselves as ‘doing IoT’, they were implementing asset tracking or fleet management or a smart grid, or any one of hundreds of other use cases. IoT is inherently vertical, and so too is AI.

It's partly for this reason that we at Transforma Insights have focused a lot of attention recently on thinking about the overall technology trends within key vertical sectors, many of which incorporate both AI and IoT, and other disruptive technologies too. The most appropriate way of thinking about enterprise requirements is in terms of the objectives that they are trying to achieve rather than the technologies that they use to do so. We have recently published a series of reports looking at ‘Digital Transformation’ (as good a term as any for the ways in which enterprises can use disruptive technologies to enact significant change within their operations and markets) within eight sectors: AgTech, Digital Supply Chain, eHealth, Green Energy Tech, Industrial Transformation, Insurtech, Smart Cities and Smart Construction, with a few more to follow. In each of those there are a set of what we term ‘Domains of Change’ where disruptive technologies are being used to transform the sector (as illustrated below). In most of those cases the Domains of Change span the use of multiple technologies, most prominently AI and IoT and often a combination of the two.

You can find more details about the research that we do on each of these vertical sectors in the Sector Focus pages.

The most critical and complex AI use cases draw on IoT

One of the most interesting conclusions from the extensive analysis of the vertical sectors and Domains of Change was the extent to which there are highly critical AI use cases that draw heavily on IoT data. While the vast majority of AI today might be focused on simple chatbots, content creation or document analysis, the more transformative impact will be from those applications that involve the use of real-time data and the analysis thereof. The orchestration of Intelligent Transport Systems, Supply Chain Optimisation or Smart Grids, for instance, all draw on large volumes of data from remote end points and process and act on it in real time.

IoT is where the ‘rubber hits the road’ as an interface between the IT world and the real world, so anywhere where AI will have an impact outside of the ICT domain will probably involve an IoT interface.

Most AI will happen on IoT devices

Back in 2022 we first unveiled our AI forecasts. We took an active decision not to focus on revenue, because it’s impossible to quantify in a way that is meaningful for all different kinds of stakeholders and it doesn’t truly reflect the impact. A valuation could cover the sum of software licence fees, the value of services or even system integration projects that have an AI component, enterprise investment in AI, or the value of cost savings. None of those really reflect the value of AI, which may just be focused on maintaining market share, and collectively they represent significant double-counting. Instead, we focused on measuring the number of AI instances. One of the most interesting findings was that an overwhelming proportion of Artificial Intelligence instances will be deployed as part of IoT applications, particularly consumer-facing products such as Audio Visual equipment (including smart speakers) and cars.

Learn more about the study here: New report from Transforma Insights predicts ten-fold growth in AI use over the next decade

AI offers an opportunity to differentiate IoT

We should note that this isn’t a one-way street. While IoT provides an absolutely critical real-time input to AI, AI is also being increasingly used to supplement and enhance IoT applications. Put simply, adding an AI element to many IoT applications represents an opportunity to differentiate because it’s hard. Take, for instance, the example of Boston Dynamics Spot robots. Building the AI algorithms associated with all the complex tasks that it performs is tough. Way tougher than developing a chatbot. Companies offering IoT solutions have the opportunity to build incumbency through AI, by virtue of developing the earliest, biggest and therefore best data set. In AI size matters and the companies with the biggest data set on which to base their algorithms with tend to dominate.

AI also offers an opportunity to completely revamp how an IoT application might be deployed, including potentially using less IoT. In a recent report ‘Video analytics as a substitute for IoT devices’ we explore how video monitoring solutions, often drawing heavily on AI, can be used as a substitute for more traditional sensors in a range of environments including security, industrial process tracking, parking space monitoring and patient tracking.? Using video analytics, businesses can unlock greater value by analysing spatial and temporal information. According to our estimates, video analysis can potentially substitute for IoT devices in around 7% of IoT applications.

Processing related to AI instances will be an increasing focus of how IoT deployments are architected

For IoT applications, which are inherently decentralised, the idea of placing increasing intelligence on the device or more proximate in some way to it (e.g. at a campus edge or network edge) is logical. Why send every last iota of data back to a centralised application when there is the option to address it locally to some degree, either analysing and filtering or even acting upon it in an automated way using AI/ML. This is inherently more efficient, reduces the load on network resources, and reduces the response time.??

The increasingly critical role will be to manage the relationship between all of these complex, specialised, distributed and powerful compute resources, in a way that optimises the application. The ability to orchestrate where IoT data storage and processing is occurring, as well as efficiently deliver data between device, edge and cloud, is a clear requirement for future IoT.

If you want to know more about this issue of architecting connectivity in the most appropriate way to deal with changing demands, including the need to manage AI resources, check out the report ‘The road to ‘Connected-by-Design’ for the Internet of Things’.

Conclusion: Strange bedfellows

Some of the hype that once infected IoT is now transferring to AI, and particularly Generative AI. But that does not reduce the potentially transformative impact that both AI and IoT can have in some of the Domains of Change that we identify. Anyone involved in IoT needs to have at least half an eye on what is happening in the AI world, because both technologies will be critical in delivering many of the use cases that enterprises need to implement (at the same time as not really realising that they’re doing IoT or AI).

And there are numerous implications for how IoT is architected and delivered. One example is the ‘Connected-by-Design’ discussion above, wherein AI will drive new approaches to orchestrating multiple levels of processing/storage from the device to the gateway to the campus edge to the network edge to the cloud. Another is the increasing requirement for AI-optimised chips on board the burgeoning number of IoT devices that will be deployed in the real world. Such deployments, of course, will often introduce many long-standing IoT constraints such as requirements for low power consumption and robustness, but now in the new context of functionally optimised AI chipsets. Considerations of further implications will be a part of Transforma Insights’ research in future.

Join us for our annual IoT forecast review

On the 13th May 2024 Transforma Insights will provide our annual snapshot of the current state and future prognosis for the IoT market, giving you the most grounded perspective on what we can expect over the next decade. We will share our assessments of market growth, technology choices, revenue opportunity, and the leading applications and sector, and much more.

We will also give our views on how AI will affect IoT. We can’t ignore AI as a driver of IoT and a factor that potentially hugely influences how IoT might be architected and enables new applications, particularly in the context of constrained connectivity. So, expect a particular focus on AI.

You can register for the webinar here: Transforma Insights’ annual global IoT market forecast update.