Fill form to unlock content
Error - something went wrong!
Get the latest updates on IoT for smart cities.
You are following this topic.
Who in the business world doesn’t wish they could predict the future? But there are some out there who try to gaze into the crystal ball—and are brave enough to publish the results. Analyst firm CCS Insight is one such soothsayer with the release of its annual IoT predictions. They’ve surveyed the whole CCS Insight staff on topics from sustainability to the future of the cloud and even the evolving definition of “IoT.” They’ve prepared a special version of the 2022 report that focuses on the IoT transformation space for readers of insight.tech.
We’ll get some of the highlights in our conversation with Martin Garner, COO and Head of IoT Research at CCS Insight. He’ll look back at 2021 to see where the crystal ball was clear—and where it was cloudy—as well as forward into the trends to watch in 2022 and beyond.
How did your predictions for 2021 play out: What went right, and what went wrong?
There were a few we were quite pleased we got right. One was that COVID would accelerate the adoption of robots, automation, and IoT across sectors. There was an initial pause in investment, but people realized they needed this stuff to keep their operations going. Another one was that security and privacy in AI and machine learning would become much stronger areas of concern. Machine learning has quite a big attack surface, and it could be initially really hard to detect a hack.
Now we did get a few predictions wrong as well. We did predict that somebody would buy Nokia, and no one did. We also predicted that the regulation of the big tech players would slow down, and it’s actually moving faster than we expected.
And then there are a few that were longer-term predictions, which we’re still waiting on. For example, a big cloud player will offer the full range of mobile network solutions by 2025. Another one is that tiny AI will move up to being 20% of all AI workloads. There is a lot going on here—especially in IoT—and the role is growing, but we’re not at that level yet.
What’s on your mind for IoT in 2022?
We have nearly a hundred predictions for 2022 and beyond, and obviously we can’t go through all those here. What we did do was a cut of those that are relevant in some way for the IoT community, and we’ve packaged that up in a report that is available as a download from insight.tech.
I’ll just highlight a few that caught my attention. Several were follow-ons from the impact of COVID. By 2025 there will be somewhat less use of office space in the developed world—down about 25%. And there will be much more use of 5G as an additional home broadband; we think maybe 10% of households will have it.
“We think #IoT is going to fade away as a term. There will be much more focus on the intelligence—the way people use it, and the value they get out of exploiting the #data they’ve got.” –Martin Garner, COO and Head of IoT Research at @CCSInsight
We also saw, coming out of last year, much higher attention paid to sustainability. We really think clean cloud is going to be something of a battlefield this year. We also think that IoT can really benefit from using sustainability in its marketing. IoT is great news for sustainability, generally speaking, and we’re mostly not making enough use of that. We also think sustainability will be built into the specifications for 6G—when we get there.
And then there’s quite a lot around IoT itself. A much greater focus on software and machine learning—a shift toward higher intelligence of things. Also a much greater linkage between smart grid and wide-area networking. We actually expect to see pan-utility—where one company is both an energy provider and a network provider, doing both by 2025, because those two networks are becoming remarkably similar.
How will the role of cloud providers such as Amazon, Google, and Microsoft evolve going forward?
One area where they’re all pushing very hard is telecoms networks. And they’re doing more in the 5G world—especially as 5G moves from its current consumer phase more into an industrial phase. If you are, say, a global automotive manufacturer and you want a 5G private network in all of your manufacturing sites across the globe, who is best placed to provide that? Well, I don’t think it’s the local telco, because they’re not global enough. It’s more likely to be your big cloud provider. So we think they’re going to become a really key distribution channel for some of the telecoms products. And I think this is a good example of where the domain between what the cloud providers do and what the telecoms guys do is going to blur quite a lot over the coming years.
Where do on-prem cloud-like experiences fit into the landscape?
What we’re seeing now is that companies like Dell, HP, and other computing providers are offering cloud-like experiences, and—this is really important—they’re offering them as a service-business model for on-premises computing. You don’t have to have the big capital costs in order to get started with quite a major computing program—you can do it all on OpEx. We’re also seeing the big cloud providers offering local cloud containers in on-premises devices—AWS Greengrass, Azure Stack, and so on—and they’re offering as-a-service hardware.
Our expectation is that on-premises will, if anything, make a bit of a comeback, and that will tend to slow the growth of public cloud. We also think that IoT is a really, really big part of this because of the strength of edge computing—the fact that we’re generating such a lot of data in industrial IoT systems, and the fact that we often need to act on that data really quickly. We can’t do everything just in the cloud; we need the on-premises side. And as IoT grows and grows and grows, we think that will enhance that trend back toward a stronger on-premises suite.
Where do you see technologies like AI, machine learning, and computer vision going?
There will be a huge focus on the intelligence, rather than on the IoT itself. What we see at the moment is that there’s a very strong focus on the tools for machine learning and AI—making it easier for ordinary engineers in ordinary companies around the world to choose algorithms, set them up for use, and build them into development. It’s actually really challenging for ordinary people to choose and use systems in this area, so we’re also expecting a lot more focus on providing finished systems for machine learning and AI. We may even increasingly see some of the finished AI bundled into things like market-ready solutions.
We are also expecting the role of smaller or specialist systems integrators to grow a lot here. They can take on a lot of the training and configuration for you, because it’s still true that the widgets that you make in your factory are not the same widgets that other people use, and you need to train the models on images of what you are doing.
There’s also a little caveat here. It’s a large task to get thousands and thousands of specialist systems integrators up to speed in this area. Maybe they originally trained as installers for surveillance systems, and they may not be very skilled in machine learning. One of our other predictions, left over from a couple of years ago, is that we will move over time toward much more distributed training, rather than centralized training. And then, having done that, you will need to trust it enough to run your operations off it.
What do you think will be the impact of making AI more trustworthy and democratized?
I think this is one of the most fascinating areas in the whole tech sector at the moment. But I want to sound just a little bit of a warning here. We think that AI is a special category of technology—small assumptions or biases introduced by a designer or an engineer at the design stage can cause huge difficulties in society. So we need more layers of support and regulation in place before we can all be comfortable that AI is being used appropriately and properly.
Another key aspect is the formation of ethics groups that are not tied to specific companies. I think we need to take away the commercial-profit focus, and instead focus purely on the ethics. It’s also clear that to build strong user trust we’re going to need a mix of other things, like external regulation. But we also then need industry best practices and standards, and we need sector-level certification of AI systems.
Then we need to certify the practitioners. There have got to be professional qualifications for people who develop AI algorithms. All these layers are being developed and introduced, but we’re just not there yet. So one prediction we have in this area is that 80% of large enterprises will formalize human oversight of their AI systems by 2024. There’s going to be a whole layer of quality control that we put in place with human oversight before we let it loose.
Tell me more about your predication of the Internet of Things becoming the Intelligence of Things.
Actually very few people buy IoT. What they do is they buy a solution to a business issue. And somewhere inside that is IoT, which is used as a technology to make it work. The real value of IoT is not in the connection that you’ve created with the things, but in how you use the data that you now have access to. With computer vision on a production line, you don’t care much about the camera; you do care about what it’s telling you.
The trouble is, we are now generating so much data that we increasingly need lots of machine learning and AI to analyze it. And then it has to be done at the edge to do it really quickly, and so on. So getting the maximum value out of those systems is going to become all about the intelligence that can be applied to the data.
Monitoring something is useful, but you still need good analytics to help you focus on the right data and not get distracted. Controlling something is even more useful—you can make huge savings by controlling things better. And, finally, with suitable intelligence, you can now optimize a machine, a system, or a whole supply chain in ways you never could before.
We think IoT is going to fade away as a term. There will be much more focus on the intelligence—the way people use it, and the value they get out of exploiting the data they’ve got. Then you will need suitable systems for aggregating and analyzing the data, data lakes analytics, digital twins, machine learning, AI, and so on. And many, many companies are already well down this path, but actually there’s still a lot to learn.
I think the ecosystem angle is a really important theme to bring out here. Very few companies can do this on their own. There’s also an interesting organizational point here for a lot of IoT suppliers. From what I can tell, most IoT suppliers are 80% engineers working on the product and 20% other—which includes HR, marketing, sales, and so on. I think it needs to be the other way around. They need to have big customer-engagement groups. If you’re in healthcare, you employ ex-nurses and ex-doctors—people who really understand what’s going on within the customer organizations, and who feed that back into the product.
Assuming you get all of that done, really a lot of the value you get comes from then applying it across the organization. And that’s a people issue more than a technology issue. It comes back to one of the truisms of digital transformation, which is that success depends on taking people with you more than on the technology that you’re using to make it all work.
To learn more about the future of IoT, listen to IoT Predictions: What to Expect in 2022 and Beyond. For the latest innovations from CCS Insight, follow them on Twitter at @ccsinsight and on LinkedIn at CCS-Insight.
This article was edited by Christina Cardoza, Senior Editor for insight.tech.