Three IoT Predictions for 2025 and Beyond

The IoT industry is undergoing major changes as mobile network operators (MNOs) confront declining profits and shifting global demands. NB-IoT, once a promising technology, is now being phased out by key players, while AIoT is gaining traction as a transformative force in the industry.

Nick Earle sits down with Ian Marsden, CTO and Co-Founder of Eseye, to dive into the trends shaping IoT in 2025. Our fifthtieth episode explores:

1. How MNOs can adapt through federated localization and smarter connectivity platforms.
2. The decline of NB-IoT and why operators are moving toward alternative technologies.
3. The rise of AIoT and its potential to transform IoT deployments.
4. Why integrating network-level intelligence is crucial for AI-driven IoT success.
5. What businesses can do now to stay ahead in a rapidly changing IoT landscape.

Tune in to hear Nick and Ian’s expert insights and get practical advice for navigating IoT’s next chapter.

Subscribe to IoT Leaders

Subscribe now

Subscribe now

Subscribe now

News

Eseye Predicts 2025 Will Redefine IoT Connectivity: MNO Evolution, NB-IoT Decline, and the Rise of AIoT

Learn more

Whitepapers Featured

Eseye’s IoT and Telecom Predictions for 2025

Learn more

Podcasts

How Amazon is Opening Doors on a Global Scale

Learn more

Blogs

Defining AIoT: Artificial Intelligence and the Internet of Things

Learn more

Transcript

Intro:

You are listening to IoT Leaders, a podcast from Eseye that shares real IoT stories from the field about digital transformation, lessons learned, success stories, and innovation strategies that work.

Nick Earle:

Hi, this is Nick Earle with the intro to our fiftieth IoT Leaders Podcast. And this time, for this one, we’re going to go back to where we started, which is where we started as Eseye as a company. And I am interviewing Ian Marsden who is our CTO and Co-Founder of Eseye. He’s just coming up to seventeen years at the company, and so who better to talk to about what the long term future of IoT could be in the industry, especially as Ian has a track record of making predictions, turning them into engineering products, implementing them, and being proven to be right. And we talk about that at the beginning of the podcast.

This is really a sort of prep podcast for a piece of collateral on our website, which is our IoT predictions report.

It’s an annual study that we do. And it is our number one downloaded piece of collateral off our website. This time, we’re going to be looking at three questions.

One is the future of the MNOs. We use the Mark Twain analogy. Are reports of their death greatly exaggerated? Or will they become the leaders in the new eUICC, eSIM, SGP.32 world? And if so, what decisions do they have to make, and what do they have to do? We have a very good conversation about that.

We then go on to look at the other end of the spectrum in many ways, narrowband IoT. And there’s been announcements on narrowband not being supported from a long-term point of view by some operators.

And we ask and debate, is there a future for narrowband IoT (NB-IoT)?

And then finally, the topic that’s on every podcast, no matter who you listen to, and it’s AI, artificial intelligence, and we really drill into, so what is the real use case for AI?

What’s the potential for AI in this world of multi operator, multi-RAT, single SKU connectivity?

Where would the value of AI really be going forward? And Ian talks about what he believes will be possible. And, again, as I said, as a person who designs and builds our solutions, it’s comforting to hear him talk about that. So that’s the podcast. And, let me hand over straight away to my discussion with Ian Marsden, our CTO and Co-Founder. Here we go.

So, Ian, welcome for the second time to the IoT Leaders Podcast.

Ian Marsden:

Thank you.

Nick Earle:

So, as you know, this podcast is an internal one. So, within the family, so to speak, it’s two people from Eseye talking to each other. But in this case, the reason is because we do annual predictions report in the industry.

So, a lot of people take a lot of interest in it from all over the industry.

And so, this year, I thought I would interview yourself. You have been our visionary for almost seventeen years now since you co-founded the company with Paul. There’s been a series of very bold decisions that you’ve taken to drive the strategy of Eseye, which in retrospect have been very smart decisions. I’m sure there’s been a few which perhaps weren’t as smart.

But most of them, I think it’s fair to say, have worked out. And they included building a federation of the operators to work in an almost airline type of surveillance model, embracing multi-IMSI very early on, bringing out a SIM that works along the lines of an eSIM with eUICC before eUICC was a thing. And then recently embracing the device resident application in front of our SMARTconnect. So, the idea of almost intelligence being within the device as opposed to being within the cloud.

So, I can’t think of a better person to interview about our predictions for the market going forward. And we do this at the end of each year, and then we find out in a year’s time how well we did, or the market tells us, if you like, how well we did. So, this year we’ve got three, and we’ll go through them one at a time just to give everyone a heads up of what the three are. The first one is the MNOs will fight to take back control of IoT.

Or if I was to paraphrase that, I’d use the Mark Twain quote, which is, are reports to the death of the MNOs greatly exaggerated?

The second one we’re going to talk about is, what is the future for narrowband IoT specifically? Because there was a lot of publicity over it. There was a lot of people asking about it, but there have also been some recent developments that starts to cause a lot of people to question it. We’re going to talk about that.

And then the third area that we’ll get into is the whole subject of artificial intelligence, which is obviously the subject of pretty much every podcast right now. But specifically, what actually is the likely role of artificial intelligence in relation to IoT? So that’s the third one we’ll get into. So that’s the table laid.

Let’s start off on the first one, and I’m going to ask you, as someone who, as I said at the beginning, has had a vision for almost seventeen years now, you built our solution around the operators and the operators’ needs. But there’s a lot of people out there that are saying, oh, well, the operator model is fading and actually the MNO model is going to take over. And they quote multiple data points, if you like, to justify that. They talk about the fact that the financial model of the MNOs is really in distress because they sell pure connectivity, but data prices have been declining.

I know in the six years I’ve been at Eseye as CEO; data prices have declined by about 10x in that period, which is quite something.

You’ve got to sell a lot more data every year just to stay flat. We know that roaming agreements break all the time, and you can’t rely on the roaming agreements. And sometimes you get quotas that get succeeded or people turn off roaming. So that is tough for the MNOs. And sometimes the big MNOs can really only cover their national market and limited roaming globally. At the same time, they’re investing a lot in building out 5G. So, there’s a lot of other priorities for them other than IoT.

And also, they’re still using platforms, which in many cases have been around for ten or fifteen years. There were platforms that were designed just to take their offerings to market. But if you look at some of the announcements and developments in the interoperability and the standards field, we now have not just the east end, but we have eUICC and now coming very fast forward since SGP.32. So, interoperability between the operators is suddenly being enabled through the standards bodies.

And if you take all of these things and you blend them together, the question is, the operators who for forty plus years have had a model of taking their offering to market are suddenly now looking at very little difficult financial situations, declining data, and a demand for multi-operator solutions, which is not what they do to collaborate with their competitors on a global basis, not just national. So that’s the background to all of this. So, the question is, what is the future for the operators? And I know you’d feel very strongly they can fight back.

And in fact, they will be the long-term winners, but only if they make certain choices and they make certain moves in the market. So, what’s your view on all of that?

Are reports of their death greatly exaggerated as the Mark Twain quote?

Ian Marsden:

So, yes. I mean, I think I mean, obviously, the thing that they have in their in their arsenal is the fact that they are the operator in a specific geographic region, and they have licenses to operate, etcetera. And, therefore, in terms of who can provide that coverage in those locations, it’s the operators. So, it’s kind of in their gift to work out how to sell it.

And you alluded on some of the changes that come out in that market, which is arguably them beginning to fight back in some of the roaming restrictions and the access fees, etcetera, coming out. And these are methods of rebalancing that shift from the roaming to the local. Now, again, you allude to historically, of course, there was more money in that connections and therefore roaming was a more expensive method of delivering IoT. And as the prices have eroded, the two have become more equitable, which is why they’re having to fight back in that respect.

But, of course, the control is with the operator. They fundamentally have the ability to set those prices. And if it’s a market that they choose that they want to fight in, rather than just being a wholesale provider, then it’s definitely for them to choose.

One thing I will point out is you might say that the price of data has come down tenfold. On the other side of that, of course, is the amount of data that IoT applications shift has increased. And if you look at applications in general, not just IoT applications, there’s a trend there that is likely to come in as well.

Nick Earle:

Yeah. In fact, just to add a little data point on that, I often own these podcasts by the way, I think this is our fiftieth podcast.

So how about that? So, you got the founder on for number fifty. But I often on these podcasts refer back to my years at Cisco where we thought for a long time, a similar dynamic occurred in the IP layer in networking.

And then suddenly voice became, it was data first, and then it became voice, which used more packets than data. And then the really big kicker was video. And when you start filling up pipes from video packets, suddenly it’s not just the price that’s coming down, but the use cases go up. And we see a similar thing in IoT, of course.

Now even basic POS devices that we have, we have a customer doing a POS rollout, sort of a point-of-sale device that is going to use five hundred megabytes a month. And you don’t need five hundred megabytes a month to check a credit card. So, a lot of these devices have video interfaces on them. But still, the pressure is relentless on the data price.

And we’ll get back to this when it comes to narrowband IoT. But it’s hard to find those value based high RPU use cases for the operators as well, for anybody, actually.

Ian Marsden:

But even if you don’t just look at the value you are seeing, the value is not just monetary value, it’s also value of the data. And therefore, I think you get a split slightly in the application space between somebody who’s just trying to buy something that’s “how cheap can I make it?” versus somebody who’s trying to actually deliver a high value service across the top of it. And therefore, there becomes a value judgment in the delivery of the service, not just price.

Nick Earle:

So, what do the operators need to do to get to that high brand of value when it comes to IoT, given the market dynamics that I’ve laid out to them?

Ian Marsden:

So, I mean, we obviously feel quite strongly about that because we, over the years, very much drove the move to using roaming to provide multi-network access for higher availability and then driving that to multi-IMSI, eUICC, to provide multi-profile access again, but increased levels of resilience and accessibility. This is all about how you provide a higher level of quality and access onto a network, which allows you to provide a better level of service and therefore hit that value end of the of the spectrum.

As I alluded to earlier, operators need to decide whether they that’s where they want to play or whether they want to play at the low end of the market, and that’s their choice. But there is clearly a value in certain application spheres, etcetera, where IoT becomes more critical and more relevant to everyday life, and things that actually we take as important, then actually that value play does come back here. So, to answer your question, operators first need to decide how they want to play. But if they want to chase that, they need to make sure that they’re looking at platforms and solutions that can actually meet those customer requirements, and those customer requirements are high levels of service, high layers of availability, both in country and internationally.

Nick Earle:

All the operators, so let’s choose it. We’re talking about an operator or a group of operators who choose the value, not the volume. Just to your point, not everyone will choose that value. But let’s say they do. They want to keep their customer relationships. They want to be seen as a value-added provider, they want to do land and expand selling in the future, because IoT has a huge amount of potential beyond connectivity.

But you mentioned platforms, they’ve all got platforms, there are some very large platforms out there. My ex-employer, I think they have about thirty-five operators using what was called Cisco, Jasper, of course, the Ericsson, what was called the Ericsson DCP, now with Aeris. And there’s a whole bunch of others. There’s a long tail of platforms, including some homemade ones. So, they all have a platform. Are those platforms being they capable of doing agnostic operator global connectivity?

Ian Marsden:

And so yeah. I mean, historically, they have they’ve not been in that sector because they were built primarily for a single operator to achieve a singular level of connectivity. And so, the risk here is that you end up with multiple platforms and making the supplier or the customer have to use multiple platforms. And historically, we used to think of this in the scenario where, you know, you’re trying to deploy in eighty countries.

You need to go and talk to eighty operators. You need to go and speak eighty languages, set up eighty contracts, etcetera. And that’s pain in its own right. But if you then add, and you wanted roaming, I now need to multiply that by three.

You know? Suddenly, this actually gets a really complex task, and, therefore, you’re looking for solutions from a platform there that can make that task simpler. And what’s interesting in a way is that at odds to just doing orchestration of profiles because that might make it you can move them between the different things, but they haven’t necessarily solved the actual platform discussion. Now, obviously, you then get some overlay platforms that would sit on top of it and say, okay, I can bring this lot together at a user interface level.

But that doesn’t necessarily bring together a support level, and that was back to alluding the are you selling premium and what is the customer actually trying to buy? So, you can we can see a shift in the market, and that that shift is moving to worrying about that overall service delivery to customers, especially those high-end customers.

Nick Earle:

So, I mentioned the federation is, you know, we have what we call our AnyNet Federation. There are sixteen operators in it currently. The people listening to this podcast that might not know what it is, could you just briefly describe what it is and specifically how it provides an architectural framework to solve that issue that you just talked about?

Ian Marsden:

Yeah. So, from the customer’s point of view, the problem I it was the problem I outlined earlier, which is customer wants to deploy in multiple, territories, multiple countries. They want to see a single SKU that can be deployed all over the world that meets all the latency regulatory support offering requirements of their applications. So that’s the sort of remit for us existing in the first place.

The question is, how do you do that in the framework of making sure that you can deliver that for long longevity, as in applications that are going to be out for twenty odd years? So that’s where the, the federation comes in because there’s a number of partners who are basically signing up to saying, actually, we are happy to enable service on the federation, and we’re happy to enable it for other partners to be able to use that. And that becomes the federation we built out, which is realizing that, like themselves, they will operate in a market. They might be dominant in their market, but that isn’t the whole world.

And if you take the reciprocal view and look the other way, it looks different, but equally kind of the same. And that comes back to your airline, alliance and now sort of scenario.

Nick Earle:

Analogy. Yeah, I was going to double click on that because and the question was going to be, which is, well, how is that different from a roaming model, which is the majority of ways people try and solve this is via their roaming agreements. But to, yeah, to shine a light on that airline model from a consumer point of view, the reason that airline lines is like Star Alliance, there’s never just one. There’s also, for instance, One World in the airline industry. The reason and others, the reason that they came about was because the customers didn’t want the inconvenience of hopping between airlines and having different contracts with different airlines, not being able to share infrastructure like gates, ticketing.

They wanted the hubbing. They wanted to be able to fly into one airport and have a guaranteed connection, ideally from the same terminal, but at least from the same airport to where they wanted to go to. So, they wanted global connectivity from a consumer point of view. And as a result of that, the big airlines went from a proprietary model to an interoperability model via an independent third-party map, in this case, Star Alliance, which is in itself a separate company managing this to be part of the airlines.

Do you believe fundamentally that a model like that is the way in which this problem has to be solved on behalf of not the other, but on behalf of the enterprise to achieve single SKU that that model has to be interconnected?

Ian Marsden:

Yes. Yes. But I’ll go back a little bit because you one of the things used to, we’ve already kind of got this a little bit in roaming. But so, I’m trying to extend that analogy. Of course, in roaming, it is a bit like in roaming, an inter-operator agreement will agree a price, so connectivity between the two operators. So that is a bit like the two airlines in the alliance agreeing what every single seat will cost for the next year.

So, there’s no flexibility in that model for them to adapt based upon capacity and stuff that’s going on in the market and whether more people are buying and what people are selling in different directions. There’s no flexibility in in their roaming pricing.

We believe that you have to get this overlay layout, which effectively allows you to do to look at IoT applications as different types of IoT applications. There’s a lot of different types of IoT applications that we already alluded to between high low end and high value, but you can do this in every dimension about, you know, are they temporary roamers, are they permanent roamers. And, therefore, the roaming model as it stands is kind of to blunt an instrument to deal with this. That’s why we believe federation model has to come in.

And it’s not just it’s a federation, but it’s also the models underneath it that allow the operators to play. And they’re happy with that.

Nick Earle:

Exactly. I was going to say that because it is not just a different way of doing roaming because the federation does offer the choice of orchestration of both roaming and localization of the connection, which is a very different model. And it’s much better for the recipient of the traffic because then they’re getting a much higher revenue per device than roaming, which typically runs at about twenty percent of the revenue per device. So, was that always a key design objective for you when you when you were building this? I mean, I know initially it was seen as just trying to get too multi-IMSI, but then the idea of federated orchestration of both roaming and localization through a central switch and a set of rules. Was that always a key design?

Ian Marsden:

Yeah. I think so. But I think it comes from the process that IoT is such a big market. One size does not fit all. And therefore, you there’s always the flex but you have to have the flexibility in the solution no matter how you look at it because, yeah, you need that flexibility in any system that’s going to adapt, especially over a long period of time, needs to be able to have that flexibility in it.

Nick Earle:

And final question before we move on to the second question around narrowband.

If I’d like the analogy you use of roaming is, like, pricing every seat in twelve months in advance, which doesn’t allow you to flex up and down for demand. I think that’s a good analogy. But does that then mean that the operators do you think they will replace the current platforms, or will this become an overlay with this capability as a native cloud SaaS pay as you go capability become a sort of clip on? Does it like a platform of platforms that sits on top of their existing platforms? Is there room for both?

Ian Marsden:

Well, I suppose we’re talking about telecoms. The world doesn’t move very fast in telecoms. So, if we’re doing predictions for next year, the answer is no. There is definitely room.

Nick Earle:

Let’s wave the magic wand and make one year tell you.

Ian Marsden:

Yeah. I think those platforms will probably get replaced, and the platforms will become more flexible because this having multiple platforms to meet multiple requirements is a I mean, not precious. They’re really good at doing that, but that doesn’t make it the right answer. And often, they do try and consolidate their platforms. And if you’re looking at ten-year horizon, then I think the consolidation between IoT platforms that are solely the in country in country rather than international ones. So, I think, naturally, they will evolve together.

Nick Earle:

Okay. So, Mark Twain got a future. They’re alive.

However, they do have to make that choice that you referred to in the beginning, which is do they want to play in the value space? Because they don’t necessarily have to. And I think that probably gives us a very nice segue into the second subject, which also has got a little bit of a Mark Twain spin to it, which is, is there a future for narrowband? Now we both know that narrowband has a huge number of advantages, technically, price, battery management, all series of advantages. And there was a massive amount of demand and interest in it over the last few years. But also, over the last few years, we’ve started to see a few announcements from some big players who said, we’ll be ceasing support for narrowband.

And so, my question then is, and there was one publicly, very recently AT&T, who are obviously a very big provider, said there wouldn’t be there wouldn’t be a long-term future of narrowband from AT&T. So, when AT&T make an announcement like that, it really is significant.

So, the question is, is there a future for narrowband? Can narrowband still be a thing?

Or is in your view the writing on the wall? And if the writing is on the wall, is that a technical issue or is it primarily a commercial issue?

Ian Marsden:

So, I think I mean, again, back to the telecoms time scales. I think in the end, the writing is probably on the wall in the long run, but equally, there are clearly some areas of the world where large deployments exist, and therefore, those deployments have got to critical mass and then what happens, say, something that isn’t going to go away and it’s fine. And so, this kind of splits into back into the previous comment about, are you talking about global deployments?

Are you talking about local deployments? Because if you’re only worrying about local deployments, then you can make the decision in the context of local ones, and that that might be a different thing. But, also, you alluded to some announcements further where, they were in education about end of life. But also, we know a lot of operators that were thinking about launching and are not being have never got there and are unlikely to ever get there.

And a lot of this is not necessarily that it’s the wrong technology. There’s lots of examples in history about good technology is not necessarily making the cut. The challenge here, historically, for many years, IoT, of course, has been subsidized indirectly by the operators because they already have the infrastructure, and therefore adding IoT onto it is a marginal cost. Now with NB-IoT, one of the challenges with narrowband always was that actually, it’s very obvious the cost, and it’s very obvious who it needs to be applied to.

And therefore, that needs to very quickly get to critical mass to be able to justify the investment or the continued investment in running it is one of the challenges you get when you can’t do across substeam.

Nick Earle:

And that helps explain why there is part of the world where it’s still alive and well and thriving, which is China, of course, where it is very high percent attributable connectivity in China. But to that last point that you just made, you have a very large operator like China Telecom who has got critical mass. So, they actually have got over that hump. And there’s a lot of money being put behind it and a huge number of devices using it.

So, it’s actually it’s got to that point where I’m assuming it’s making money for them. They’re really, really pushing it. But it’s difficult when you then follow the long tail of operators down to the rest of the operators where it tends to have three or four operators per country and one hundred and ninety countries. Is it tech or is it commercial?

You could argue it’s actually to do with the fundamental economics of how to make money from narrowband.

Ian Marsden:

Yeah. And I think so. And then you the global element also didn’t help. So, I think if that’s just if you look at it within the in the regional.

But if you look at the global element, the problem and historically, you know, why 2G still is even talked about in this industry is because of the need the globe is a big place and lots of operators, and they will have different networks at different stages of evolution. And, therefore, the need for backwards compatibility is one of those things. If you’re looking for global deployments, it’s really important. And the problem with narrowband, again, needing different modems, and you look at the modem, some of them have compatibility options.

But if the only reason you were designing it is because you need the power saving that narrowband would give you, you then couldn’t make use of the fallback options. It all adds to the complexity and the problem that means that potentially the business case for the customer doesn’t stack up, let alone for the operator.

Advert:

Have an IoT story to tell? We’d love to feature you on the podcast. Email us at iotleaders@eseye.com to apply and share your experience.

Nick Earle:

So, if I was a customer, what advice would you give me advice at home?

I was about to bet the farm on narrowband IoT. Actually, a lot of agriculture use cases are going with narrowband IoT. I was going to bet the farm on it. What are my alternatives if I want to protect myself going forward?

Ian Marsden:

Well, first up, we’ve always the question about where you’re going to deploy it. But I mean, it’s interesting because we obviously help customers with SMARTconnect, etcetera, build a number of different products in there. And often, you have to look at not the technology could be an option to help you, but you have to look at what the alternatives are. So, I think if narrowband is not available, what it can do?

And that’s about your power profile, your power budget. You know, are you making use of alternative energy sources? And you also have to look at often, sometimes customers look at the modems and go, oh, that modems cheap. But you have to factor in, is it cheap because it’s a good service over the lifetime of the product.

If that product’s a twenty-year lifetime that’s much more significant than necessary saving a small amount in the modem. So, to a certain extent, you have to look at the whole business opportunity holistically, my viewpoint on that.

And, again, we have customers that we help design their products. Sometimes you do end up with almost a counterintuitive decision because you have a truly global product that might not be sending a lot of data. The category of modem ends up going because you want long lit long life maximum coverage in all over the world. And, actually, that might drive a different decision.

So, there isn’t there isn’t one answer because if you’re buying building a small low cost, low lifetime products, you might possibly would make a complete decision to something that’s going to be out there for twenty years.

Nick Earle:

Okay. One other question some people would say, well, with satellite coming, one of the issues in narrowband, as you pointed out, is the just the investment versus the return.

But with satellites, of course, you get truly global coverage, oceans, land, agriculture. It doesn’t matter if you have farms, remote farms. It’s no longer the towers or where the people are. The people tend to be around the cities, but a lot of the use cases are everywhere else. But there’s, therefore, satellites over the long term will solve this. But does it really solve it because if the underlying operators aren’t supporting it because they can’t make money?

Do satellites solve the narrowband conundrum?

Ian Marsden:

Well, satellite solves a different problem, but it doesn’t actually solve the physics engineering part of the problem because of the power consumption.

But then if it kind of comes down to what’s the problem you’re trying to solve, because satellite coverage in emergency situations or very high value situations is great.

Nick Earle:

Yeah.

Ian Marsden:

So, and of course, there’s what’s interesting equally is on a lot of applications, the value of the application actually changes dynamically, as in something that’s monitoring something that’s in a normal condition is very low value data.

Nick Earle:

Yeah.

Ian Marsden:

If it becomes, like, you know, something’s been stolen or something and you need to track it, then that becomes very high value data.

So, and that’s not only in terms of power consumption, but that’s in terms of actually value of the information that’s being delivered. And therefore, you can so I think multi-RAT in that element, I think, whether it’s satellites or these RAT types is all part of that mix, but it’s down to the application of the business case.

Nick Earle:

Okay. Application. Okay. That’s a good prompt to move to question number three. I can’t think of a way of bringing Mark Twain into this one.

Who knows? Maybe Mark Twain was an expert on artificial intelligence without knowing it. AI is everywhere, we all know that.

It used to be cloud washing and is very definitely now AI washing. Everyone now has an AI strategy magically.

But in IoT, I don’t believe that there’s really been a clear definition of what role AI will play. I know you’ve thought a lot about this and how in view, and, from what I know of that, it’s not a view that I’ve seen elsewhere. So, I think it’s worth exploring.

So basically, this is an open question, which is what role do you believe AI could play in the future? And don’t worry about the one-year timer. Let us give you two or three years. What role could AI play? Where is the value in IoT for an AI use case score? What’s your view on this?

Ian Marsden:

Well, so I think where we sit is a slightly different question because in a way, IoT is a number of players coming together to deliver an overall solution. If you look at any of our applications, there’s always multiple stakeholders. And in fact, that’s one of the reasons that IoT is complicated and difficult to put together. But there’s a number of stakeholders in there, whether it’s the end customer, the enterprise that’s selling it, whether it’s the network provider, whether it’s the cloud system, the back-end provider, or whatever.

All of those have pieces. They all say they work in IoT, and they all deliver IoT. But actually, therefore, where AI comes into that, though, is going to be different if you ask each of those players. Now, obviously, we’re one of those players in there.

But I suppose if I was looking at where it comes in first, it clearly comes in first at the application there, which is alluding coming from your previous question. And the reason is one of the issues is about repeatability and auditability around those elements. And at the moment, in what a lot of them, the network elements need the repeatability and auditability for those elements. It’s a bit like, you know, safety testing and those sorts of things, the closed loop algorithms, etcetera.

You need to make sure the loop is closed and is under control. And so, the networking side, I think, is probably for at least the immediate year is doing machine learning control level of closed loop. So where does AI come into this, which is clearly being used and coming in, we see an increase in our number of customers that are using it on their application side. And whether that’s in the back end of the application for making better decisions or around the application itself.

Or actually on the device side as well. We’re trying to process large amounts of device on the device and therefore remove the need to send all of that data straight up into the cloud for further processing and therefore be able to make decisions on the devices.

Nick Earle:

You mentioned network level and network layer. Do you see that as being distinct from the application layer? In other words, is there a an untapped opportunity for AI to provide intelligence insight from rich network layer data, which goes back to the first question of the MNOs and if they could embrace a multi-operator environment where every device has a different network state, if any other deployment, is there an opportunity to provide network level artificial intelligence.

Ian Marsden:

Like APIs, for example. I think it will evolve through the normal processes because the controls and the guardrails need to be put in place to make sure that that all the stakeholders are happy for that level of control. But I think you’re clearly, if you look at a federation where you’ve got multiple players and multiple routes to be able to deliver connectivity to a device, the decision about which route to being able to use is not a simple decision. It’s based upon lots and lots of factors, quality of service, requirements of device, location, availability of networks, etcetera. And this is all information that today is being processed by algorithms inside of people like us.

But that should sort of take those guardrails off and pass it over to a high-end agent is coming, but it’s not necessarily there. But even when you do it, you still need to define the definition of success, as in what are you trying to achieve by this? Because not every IoT application has a different definition of success because they’re trying to achieve different things by their connectivity.

Video streaming application is completely different to a media application, for example, and what success looks like is different.

And what that means is that those algorithms need to be controlled, and those data rates need to be configurable by APIs. And that that kind of is where we are focusing. Our attention is making sure that those APIs do define the rules the network is going to work within is the next step of that integration because then the customer or the, federation partners can define the rules, and it’s that conjunction of rules that means that manages those devices. Because, you know, you’ve got millions of devices on network, you’re not managing them individually. The machine has to be managed by the system.

Nick Earle:

This has come up actually on a previous IoT Leaders Podcast, but we didn’t actually position it with an AI wrapper, and it was the one that Amazon did talking about their Key for Business project, which I described as a small ambition to own every door in the world. But essentially the idea of automated entry to any door by any delivery company, by having secure access and the ability to get inside and have it recorded and use the consumer and see it’s been done.

So, you don’t have to wait in all day. And from their point of view, they increased their first time percentage of Amazon Prime from eighty four percent to ninety two percent, which is worth billions of dollars. So, we talked about that in in a podcast and listeners who want to know more about that, you can find that by searching through the list of podcasts. But one of the things that the, the main guy, the project owner, Vijay said there was right towards the end of the podcast, he said, right now we’re doing optimization of the connectivity down at the ZIP code level, but ultimately, who wants to do optimization of connectivity down at the device level?

Because there are now, with an agnostic platform in the way we described in question number one, a platform that’s agnostic to the operator, you now have multiple paths to connectivity cellular connectivity path, paths to every device. And if it is agnostic RAT, it’s going to get multiple RAT types or even more paths to every device.

As you pointed out at the beginning of your answer, it’s not just, oh, do I have connectivity? I’m going optimize it as the MVNO, and I’m going to choose the one where I have to pay the least because I’ve got a fixed price contract to my customers. I’m doing least cost routing to the device. But now it is the network state information of like, what is the network like?

What is the latency like? What’s the signal strength? Like a whole series of network parameters, which actually are ingredients in making the decision on the switching algorithm, as you call it. So as the number of right now, we’re not making granular switching.

The industry isn’t making granular switching decisions based on rich network data at the device level. And what you’re saying is that once you can cap out that net all the devices on one network, in our case, it’s one MPLS network, from sixteen operators and almost eight hundred roaming agreements from millions of devices. But once you’ve got the network data, then you’ve actually got the data that can be used in machine learning or as training algorithms or whatever, training data to actually have AI decisions to, in my mind, achieve the Amazon vision of having long tail personalization of the connections where they are dynamically choosing the best condition, best connection, or best network connection based on exactly what is happening right now with that device, which might be different to what’s happening to that device in two hours’ time.

Ian Marsden:

Correct. And that that leads to a point of also, the one of our focuses is about getting consistency from our partner networks so that that data that’s being interpreted and processed, no matter which partner you’re using or where you where your device is in the world, is has the same level of provenance so that you can actually make those decisions on it. Because if you’ve got a weak link in that chain, then that becomes something that that could drive a wrong decision.

Nick Earle:

It’s almost like a metadata abstraction of if we’re going to make decisions based on thirty parameters, not four, three, or five, then you’ve got to be able to collect that data from everybody, which comes back to how you interconnect with those. It’s got to be right in the foundations of your interconnect with the operator, because otherwise, you’re comparing apples and oranges, and you don’t know what it all means.

Ian Marsden:

Correct. Absolutely. So, I need to bring you back to that question. That’s where we see that that going, and that’s really what we’re focusing on from our point of view to, you know, meet the Amazon vision, etcetera.

But, equally, in the short term I’ll have to pull us back slightly. In the short term, we see increased use of AIoT really with on the custom side, either where that’s video processing on the device or in the router devices, or even on the cloud. So that’s really where those entry points are from that we’re enabling our customers to do. And we can help that by, again, providing them those APIs and those data feeds, for rich data that augments that.

But really, the evolution from our point of view is it coming across every part of the IoT chain, not just the applications.

Nick Earle:

So which brings us in a nice, neat circle right back to where we started from because we talked about the need for the operators to not only embrace new platforms for, I’m going to use you and that was dynamic pricing, but also to do land and expand selling because simply doing connectivity is not enough. They want to climb the value stack if they’ve chosen that as a strategy. And one of the ways they can climb the value stack is if there was a way of getting this rich network level data to optimize their models and their value-added services that they offer to their customers.

So that is probably going to be a very big subject that we will go into a lot more detail in the future, but is a is a real use case already backed up by several podcasts that we’ve done already of people saying the more advanced users, they were really deploying very large, vast global estates and have a lot of technical capabilities like Amazon and others, saying, I want to go down to this area because the value for my use case, in Amazon’s case, the cost of not delivering a parcel is one hundred times greater than the cost of providing connectivity to the device. The price is so big to get your quality level up that they and they know they need to use analytics and advanced algorithms and what we would have called this AI, network level AI APIs to be able to do that.

Because if they can get eighty four percent to ninety two percent on fifty billion packages a year, whatever it is they do, and I don’t think that big is too crazy. I think it’s not too far off. The cost of that and the benefit to Amazon is huge. But the benefit for us as consumers is even bigger knowing that we’ve got we talk about trust and providence.

We just know that the service is going to be delivered in a way where AI will be taken care of a lot of the choices providing you’ve got that rich network level data from all the different operators, which is why it has to be some form of, we would say, Star Alliance network infrastructure, where you can see all of the all of the operators and get that data from all of them interpreting in the same way it’s sent. So that’s what we say. Let’s draw an end to it there in terms of the predictions. I think this prediction podcast is different to the others, not just because you’re on it this year, or because all the Mark Twain references, but because I think the predictions that we last year, it was all about the device.

The theme was every prediction we were saying was everyone pointed back to the device. This year, what’s interesting is that these are longer term predictions.

It’s not like something’s going to miraculously happen in 2025 but the trends appear inevitable that over a longer period of time, these things are very likely to happen, which is exactly how you started sixteen, seventeen years ago of predicting things and then seeing them play out over a long period of time. So, you do have form in terms of being able to predict these and we will see in the medium term, shall we say, how these things play out. But in the meantime, Ian, thanks very much for being my guest. Thanks for being the fiftieth podcast and second appearance.

And thanks to our, listeners and viewers. I hope you enjoyed this. The Eseye IoT predictions report will be available on our website. And as I said at the beginning, it is I think I’m right in saying it’s in a more downloaded piece of collateral so I am sure there will be a lot of interest in it. So, Ian, as a CTO and founder of Eseye, thanks very much for being on the podcast.

Outro:

You’ve been listening to IoT leaders, featuring top digitization leadership on the front lines of IoT.

We hope today’s episode has sparked new ideas and inspired your IoT and digital transformation plans. Thanks for listening. Until next time.