Data & Dimensions | Ep. 3 - Digital Twins In the Industrial Metaverse

View Full Transcript

Neil Barbour   00:00:00

Hi I'm Neil Barbour, Industry analyst with S&P Global Market Intelligence Kagan. Welcome to Data and Dimensions, a podcast dedicated to exploring the transformative evolution of the Internet into immersive 3D spaces. 
 

Ian Hughes   00:00:10

I'm Ian Hughes, analyst with 451 Research and I've specialized in emerging tech for the majority of my 35-year career. I primarily write and present about the rise of industrial metaverse use cases as the multibillion dollar games industry meets the multitrillion dollar operational technology world.  
 

Neil Barbour   00:00:35

Today, we are going to be talking about digital twins, one of the buzzwords often used across this stack of applications and particularly in the industrial context. It's very much in your wheelhouse. 
 

Ian Hughes   00:00:46

It certainly is having covered industrial IoT for the past decade. Industrial IoT was all about gathering the information in the first place from plant machinery and general industrial processes. But all that data had to go somewhere. And really, digital twin is about collecting all that data contextually in one place so you can see what's going on with your plan. But to many people, they hear digital twin and maybe only think about the visual side of twin, the 3D model of it. And it's both. It's where something is, but it's also the current state of that thing. And that's why I guess it can be a little bit confusing for people, even though it's a really handy buzzword to hook on to things because people go digital twin, I get that. I understand that. 
 

Neil Barbour   00:01:37

I fall into that trap of where is your digital twin? That's a 3D asset that represents something in the real world, but it falls into a broader context. In fact, if you were to just think of 3D assets, I mean, we've had representing 3D assets representing something that exists in the real world, we've had that for a very long time. That's computer-aided drafting or simply something that shows up in an unreal engine. But you're talking about pulling across those sort of simplified assets into a broader ecosystem. 
 

Ian Hughes   00:02:10

Yes. And it's understanding the state of something. If you take a pure view of a digital twin. It's saying there's a physical thing and it needs a twin. That's what the words therefore. So it's saying, show me everything about the thing. So it's where is the process, what's happening, what the machine is doing, what's the temperature of one, how fast is another running. All that stuff is about live information that you would otherwise have to go around and maybe look at each individual dial and meter on a device across a large property of a factory. And so these things kind of exist already because you have control boards and you have centers in factories that know what's going on. So they already are a digital twin in that respect. But the difference here is, can you make that even richer? Can you add information to things like the 3D model so that when somebody is out on the shop floor, they know where something is by what it looks like in -- through an AR headset similar, just to make things more efficient. So if you got all the information, you might as well kind of use that. And we try and ask people in our industrial metaverse survey about digital twins and quite common survey questions that people will ask is, do you use digital twins yes or no? So we ask that, but that's not really -- that's like a screener question for are you engaged in the evolution? 
 

Neil Barbour   00:03:47

Yes. "Do you understand what this is?" 
 

Ian Hughes   00:03:49

Yes. So in our survey, in our 2025 industrial metaverse survey of 863 people that were engaged in industrial metaverse, 74% say just say yes straightaway to digital twin. So it's a thing that they understand and the building. But then we go on to. 
 

Neil Barbour   00:04:12

Do you have the year-over-year comparison, handy? I mean, 74% this year up from. 
 

Ian Hughes   00:04:17

It tends to be -- the last 2 or 3 years, it tends to have been pretty much the same. It's gradually increased, but not a huge thing because people have been engaged in this for a while now, and it's --... 
 

Neil Barbour   00:04:29

Yes. those consistent returns almost reinforce the idea that this is an established methodology or approach within the sector. 
 

Ian Hughes   00:04:38

Yes. And as a concept, everyone's bought into it because it's easier to think about digital twin than it is to talk about the specifics of some weird industrial IoT sensor or some unusual protocol that's moving some data around. So it's a much more C-suite friendly term to say digital twin. And so I'll generally say yes, because a lot of these industries are gathering more and more digital data than certainly much more than they were a decade ago. 
 

Neil Barbour   00:05:10

I thought that was an interesting point I just wanted to swing back to is that the digital twin to some degree, already exists inside of these factories, but sort of in an abstract, data-rich visualization light respect. And so the other part of this digital twin is adding the 3D spaces to it so that you can interact in a more intuitive way. maybe remote assistance and maintenance in the plant is easier to come by. It's easier to see the way that maybe certain dynamics within the plant are impacting other parts of the plant, things like that. I just wanted to underline that case because it sort of set my brain a light there. 
 

Ian Hughes   00:05:50

Well, it's one of the more kind of -- it seems to be obvious use cases for the data to be able to play it back to the shop floor, so somebody can know where something is and what's happening with it. If that's easy to do because you're already collecting all the information and all the data anyway, then it's not a huge step to say, well, present that information to them locally. So rather than just a big board in the control room with a load of lights on saying everything is, you take the board with you relatively, but the board is even richer. And one of the examples that somebody cited at one of the conferences, it was one of the vendors years and years ago, they said, well, we've got this model of this very big refinery or refinery or something. And somebody has to go and fix widget #37, whatever. And they need to know what they need to take with them to fix it. And so there's the basis of you need a spanner and you need to talk wrench or whatever to deal with it. But where it is, you might need a ladder. So if you go wandering out to it and you don't know where these things and it's higher up than you can reach, then you need a ladder. And so telling somebody you need a ladder before you go is quite useful. I mean it's a very simple example, but -- that's illustrative of the -- yes, the larger ROI to be reaped here or harvesting the idea of saving worker time and giving them a better visualization of what they're running into. And it was another example with trains when they're servicing trains, having a large cart with all the tools on the platform that they need to do the fix to be in the same place as where the thing needs fixing is when the train comes in. So the location rather than having to shuffle all the way up and down the platform, trying to find where you can fix it, have it indicated where that is. So location becomes incredibly important. So that's why I mean sometimes people get again freaked out by there's a whole 3D model of things and all these elements going on, it looks like a game or something. We have this chatter with all the metaverse stuff, don't we. But the notion of tying in relevant information, some of which is spatial and some of which is just operational like here's the right size screws you're going to need to put to fix this. All that makes work more efficient, and that's going to save money on in operational terms for whatever industry you're in. 
 

Neil Barbour   00:08:33

The survey you guys feel, it's saying there's strong uptake or there's strong familiarization with the terminology. What else is it saying? Spool that narrative out a little bit for me. 
 

Ian Hughes   00:08:43

Well, we try and drill down to say, okay, you've said digital twins, what type of twin are you using because there's not -- there's not one answer to twin. That's again, a common misconception. So we ask these various areas. So we say, what do you use it for? And I'll just list them. So there's testing and certification, conceptualization and design, operations and maintenance, manufacturing and assembly, training and remote support, customer and aftersales service, end of life and recycling. So you can see in there, we've covered a whole range of things that these twins may be important for. I mean they're important for all of them, but which ones are the people hitting first. And quite often, and this most recent survey, our 25 survey had about 58% were using testing and certification digital twins. So that was for simulations of things and checking things were going to work. 
 

Neil Barbour   00:09:51

Right. So do you get the impression that is largely when you're planning a plant or an expansion inside the plant, you can test the -- how it would respond to certain external forces, stuff like that. 
 

Ian Hughes   00:10:04

So that's the elements of the physical world, what's it going to do to this stuff that you're trying to make. But the other one, that's about 52% response because people have more than one twin is conceptualization design. So that's the actual first part of design to say, let's effectively be digital first. And this is a quite confusing digital twin because this is a twin of something that doesn't exist yet. So this is -- it's kind of almost like the wrong terminology for it, but it fits. So it's the start of the digital thread. So you build and simulate the things that you're going to be doing in your factory and even the factory itself completely digitally. And it's in a -- it can be in a full 3D representation of it or it can be just some underlying simulation. But let's go with a full 3D simulation of it because that's. 
 

Neil Barbour   00:11:02

Do you get the impression when people are building out what may have been known as just computer-aided drafting before. Is that separate from what they're doing with when they design a digital twin? Or in other words, is the production more complex when building a digital twin than rather just an architectural draft of a plant? I assume the answer is yes, but what more color do you have? 
 

Ian Hughes   00:11:27

That's one of those yes and no question answers there. So traditional computer-aided design is precisely that. It's using the computer to do the drafting and all the things where you're going to put everything and work out maybe some of the tolerances and what fits where and generate a bill of materials for the thing that you're going to end up building. But when you take that design, if you say that, that's just a design, say, like an architectural type design, it may have some motion to it or where the COGS go and things like that. But when you get a bit deeper and then say, right, well, we're going to take that structure, that thing, whatever it is, whether it's an entire factory or a car or whatever, and we're going to simulate it completely digitally in this -- in an environment to work out how it actually works and what rubs on what and what bumps into what. That then becomes a much richer version of CAD. So that becomes... 
 

Neil Barbour   00:12:30

Right. Merging those modalities into one file type or. 
 

Ian Hughes   00:12:34

Yes. And you're then saying that the benefit from that is you've simulated all this everything about something. And then you can then start building the physical thing. And as you build the physical thing, this digital model you have of it in digital twin then becomes effectively the operational digital twin of that factory or facility because it's a one-to-one relationship. And so it becomes the thing that then receives the live information from the physical world. So you've got virtual made up stuff, simulating what may or may not happen. And then as you transition into running it, you've then got an operational digital twin to run with. And they come from the same place, and they are linked and they are part of that digital thread. 
 

Neil Barbour   00:13:25

So what else is on the list? What's third? 
 

Ian Hughes   00:13:28

Well, third is operations and maintenance. So it's actually running the plant. I think you kind of get down to about 40% they're saying they're using it for training and remote support of technicians. And those are -- we've kind of bolted 2 things together there because you can only ask so many questions at the same time of people. But the remote support element of being able to say, help somebody on the shop floor, here's all the stuff that you need, whether it's the latter example I just -- I gave earlier. That's one thing. But the -- being able to train people in the virtual environment of a factory before the factory is built is a really beneficial one. And that's one that's kind of -- we're really on the cusp of proper industrial metaverse here to say, well, here's a virtual factory, all go into the virtual factory and learn what you need to do and make adjustments accordingly. And then when the real factory is there, just walk into the factory and carry on as you were doing. in their. And that's the -- I'm not going to say dream, that's the direction that hopefully, the industry would go because that saves a lot of time and effort and it's safer and quicker and you're just operational straight away. And the other important element of that, which is in other parts of the survey, though, is that you use that environment to also train autonomous devices and robots. So you use the virtual environment to allow -- because it's a virtual environment as accurate as possible in the factory or facility you're going to build. And then you throw your robots in and let them wand around this virtual environment, doing the job they need to do, bumping into things, not bumping into things, working on the best route around things. And you can... 
 

Neil Barbour   00:15:25

Making the mistakes in the low stakes environment. 
 

Ian Hughes   00:15:27

Absolutely and making hundreds of thousands of mistakes over what would be otherwise many, many, many hours, but almost in minutes. So you ramp up the training of these things. And then when you open the factory, you take the brain that's been trained, the autonomous brain, some companies call them brains. -- take brain and you drop it in the physical robot, it doesn't know any different. The sensors that it's using, the things that it's picking up on are exactly the same. So there's no training time required potentially. 
 

Neil Barbour   00:16:04

Yes, it's been in the matrix. 
 

Ian Hughes   00:16:06

So they are... 
 

Neil Barbour   00:16:08

It's been a brain in a jar. 
 

Ian Hughes   00:16:10

Yes. So robots born in the metaverse, whereas we kind of are born in the physical world and try to be in the metaverse. And except for us, it's more jarring because we go, oh, this is different. Obviously, they're not senting beings. But for them, it's no different, and they don't care. 
 

Neil Barbour   00:16:28

It's interesting. The more as we go to these shows, we see these robots that can walk around and articulate and find themselves in a space. But the thing that's probably lacking like driverless cars, the more data you have, the better the driverless car experience can be. And in most cases, you need data down to the millimeter or to the enter to a very small quantity of mapping that entire city out. So you reduce the mistakes to hopefully a very low level. It's just like in a factory, if you can give the robot the map and let it drive around and make all the mistakes virtually, then when you drop it into the real world, it's going to just be ready to go. 
 

Ian Hughes   00:17:09

Absolutely. And it's the amount of time and money that, that saves that is almost a no-brainer, I would say. So if you think, well, I've bought this robot and it needs -- it can do all sorts of things. Maybe just a trendling round ATV, little truck going around doing things. It's got a certain amount of in-built autonomy. It knows when to stop some things in the way and things like that. But as you said, it doesn't know the map. So it either has to go and find it maybe like a vacuum cleaner bounces around your house until it's worked out where the walls are or you tell it in some way by giving a boundary. But here, you're giving it absolutely everything and trying and letting multiple robots interact with one another as well potentially and with the people that are also being trained in that environment. So we have a fully operational factory, but digitally. 
 

Neil Barbour   00:18:12

Yes. And it underlines the importance of having a really good digital twin to stand up that... 
 

Ian Hughes   00:18:17

And that's the challenge at the moment is that there's the kind of marketing side of things and saying we'll do this. And we all know that it is incredibly difficult to 100% simulate the physical world and get everything right. But it's getting closer and getting better. And it's getting easier for these things to do. And even if you do it a little bit, even if you're training 80% of the robot situations, then it's still cheaper to have a digital version of the factory than it is to build a factory and then wait for 6 months until you can use it. 
 

Neil Barbour   00:18:57

So what are some of the technologies that are advancing the creation of digital twins? Is this where we talk about the AI story? Or why is this becoming more of a reality now? 
 

Ian Hughes   00:19:09

I think the initial challenge has been that with the rise of this instrumentation and everything being instrumented and everything spanning out digital feeds of data potentially, there is a whole massive data and information being generated at huge rates from any kind of mechanical device anywhere. It doesn't mean everyone is doing it, but there's the potential for that. And so being able to manage that and understand where that is and what that -- what you can do with that has said, well, let's contextualize this data and so we know what it means at a particular point in time. So that's become the reason for this core data structure of a digital twin to come into existence. But then... 
 

Neil Barbour   00:19:59

Interesting. So you're saying as the IoT era ramped and the sensors everywhere heuristic took hold, we just had more data to work with, and now we're sort of organizing that data... 
 

Ian Hughes   00:20:10

Yes. And so those twins and twins within twins and it's twins all the way up and all the way down, that becomes the underlying, say, data architecture for those things. And then that is an incredibly good feed for AI. So the fact that this isn't about LLMs writing power tree or anything like that. This is about the things that have been happening in the industrial world around predictive maintenance and general machine learning on -- right on the shop floor. This is being improved further by all the other advances in AI. And AI needs data, but it needs data that makes sense. And it can find weird anomalies and things, but it needs good structure and people often, again, in some of our other surveys will get asked are you doing X, Y, Z and they go, well, we're doing X, but we're not doing Y and Z. And they say, well, and are you going to use AI? Yes, we'll do it to do Y and Z because we're not doing them. And AI is not going to be able to do it until you've got the rest of the things sorted out. And so we often get that's a common problem across all enterprise AI users. But the volume of real-time data and the volume of historic data as well to be able to draw upon to say, how do we make this process run more efficiently or more safely and more efficiently has been a draw for digital twin because you can say, let's get the twin to do it. The same as I was describing about training autonomous robots as well to say, well, we've had simulations. Every industry is full of simulations, and that always has been. But it's tying all those simulations together into a unified thing. So again, it's contextualizing a multitude of simulation into one understandable thing that you can, a, train robots in or b, train people in. And so it's all part of everything kind of coming together at the same time, I think. 
 

Neil Barbour   00:22:20

Right. You mentioned that AI needs data, but in some respects, we need AI to sort of structure this data and then allow us to push in and pull out the necessary insights to make that data to make that data usable. So what else is the survey saying as we go down the list of questions? 
 

Ian Hughes   00:22:39

Well, the next part down this path is not just which twins have you got, but if you have a load of digital twins doing lots of different things, are they integrated with one another at all? And this sounds like it should be obvious that if you're advanced enough to be running those digital twins that they'll all be kind of sharing information or part of a unified process. But the nature of these things is that quite often something gets dropped into a process. It has a nice digital twin for it, but it's not integrated. And you end up with with a digital silo as opposed to an analog silo that you may have had before saying, well, we only know about this machine in this context at this point in time. We don't know where that sits in the whole supply chain issue or what's the impact going to be if we change this or some of the inputs to this machine. And so we ask how integrated are you? Or do you consider your digital twins to be? And again, there's no easy answer to this. This is a huge architectural issue. But there's an operational integration where twins are maybe sharing -- doing some real-time sharing of data. So you got 2 machines on the line, you're aggregating the data together. Maybe one tells the other you're going too fast, you're going too slow or hold up. Those kind of things. So that's the sort of operational integration. And there's also people responding with partial integration. So they've got some of that and a little bit of life cycle stuff going on. And they're the kind of top 2 responses. So it's like 23% and 18% of respondents are in that. And then there's people that are exploring integration, they account for 15%. So they're not yet doing it. But then there are some more advanced things such as having a full ecosystem-wide integration and separately a life cycle integration. So is there a full digital thread going on between all these various pieces of information you have and data you have? Is that process pulling together? And are you getting the most from it? And then kind of almost -- it's not really separate, but it's a different thing about having digital twins of your product. So if we're talking about making stuff, you've got -- we've just been talking about digital twins of factories and everything in the factory. But when -- if you're making a car and then the car is out on the road, a customer now has it, are you gathering information from that car to help understand and support your customer and then maybe aggregate all the stuff together from all the cars that you're getting all this digital information from into a huge digital twin. 
 

Neil Barbour   00:25:43

The galaxy-brain car. 
 

Ian Hughes   00:25:46

So that kind of customer integration there is about 12% we're saying that they're engaged in that. And then there's only 7% that say that they don't, that everything is stand-alone. So there is obviously a need -- I mean, a need for integration. And this is 1,000 plus -- 1,026 respondents in both industrial and kind of enterprise and some B2C space. But so it's giving us a good view of the fact that they didn't all go, oh, no, nothing is integrated, which would have been shocking. And -- but it does show the complexity and the difficulty of understanding the maturity of anyone's kind of entire digital transformation. But a digital twin is handy for that, I think. 
 

Neil Barbour   00:26:44

What do you think is slowing a deeper integration of the digital twin stack? Is it CapEx issues? Is it just the standards don't exist? Is there sort of just a lack of clarity about the benefits? 
 

Ian Hughes   00:26:58

The standards don't exist. There's not -- there are consortiums looking at digital twin standards and things like that. But this is -- if you're talking about it, particularly as we are here in an industrial context, the merge between operational technology, the people that make the machinery run and the IT people is still a little bit of a battle going on as to who's in charge and who does what. And IT tends to go, "Oh, we'll just patch this and let's install another version of this server somewhere. And that's what IT does, does manage it, but it's about replacing everything. And the OT world is no, this stuff is running. We can't turn it off. This is our business. We need to keep it going. And by the way, that fancy thing you're saying, I don't see the value of it, just we don't do technology for technology's sake. So there's practicalities of differences in opinion. IT, we do tend to be, that's shiny, let's have one of those. And OT is we'll have one of those when we know it works. And there's a delicate balance there and different industries are slower sometimes than others in picking those things up. But there is a definite -- see there's a definite understanding that things are moving on and 3/4 from doing digital twins in some form. But that's taken decades from the past 10 years of industrial IoT watching this happen at the start, it was like this is what should be happening. And we're only 30% of the way there towards the kind of more full industrial metaverse for most people, I would say. 
 

Neil Barbour   00:28:53

What are some of the industries that are most exposed to these technologies? 
 

Ian Hughes   00:28:57

Well, quite often, discrete manufacturing of some sort, it tends to be more -- tends to pick things up quicker because there are more greenfield sites being built to make something. So constantly -- if you're dealing with the process industry, maybe like oil and gas or paper and pulp or chemicals, whatever, you've got a plant that's maybe been running for 20 or 30 years. You do apply new things to it, but it becomes -- it's a core asset and it's already working. So it's tweaks. Whereas if you're saying we're going to build a brand-new car plant, then you can go right, what can we use now to make that easier to build that and then run it. And so you can adopt that technology quicker. And for instance, like a digital twin of a car is a lot easier than a digital twin of a barrel of oil. 
 

Neil Barbour   00:29:57

So what else is the survey telling us? 
 

Ian Hughes   00:29:59

Well, I think it's continuing to say that there is an evolution. There's not a single point where you go, yes, I've achieved a full digital transformation. But I think the key is the state of the digital twins and where you draw the boundaries, obviously, is slightly variable, but the state of digital twins and their integration across an entire process is kind of a maturity index for the acceptability of -- or the ability to adopt some of the new technology and be able to slot some of those things in. And so when we look at things and when vendors are looking at things, they can look at our surveys and kind of see where they should be aiming and what they should be saying and what the kind of pitch should be for certain types of industry because these things break down. As I said, it's different in discrete to continuous. It's different in auto manufacturer to kind of electrical equipment, for instance. So those kind of breakdowns and just saying, well, this one size fits all thing isn't going to work yet. I mean and industry, as I've said so many times, is full of -- you can have a steam engine at one end of the plant and a quantum computer at the other. And it's just all running at the same time. So it's like everything is there. And what gets improved first and how it gets improved and where it gets improved and is it an improvement, but it's going to take a long while. But the concept of going IoT, digital twin, put it in -- make it part of a whole digital thread, pile all the technologies together, you end up with an industrial metaverse, I think, is a good top-level thing to kind of consider. 
 

Neil Barbour   00:32:04

Yes, a really striking concept, born of efficiency and ripe for probably some substantive ROI over the long term. I want to circle back to training employees on in a digital factory that maybe doesn't exist yet. Is this where XR hardware comes into play? I mean sometimes when I go to a trade show and I see a demo of AR assisted maintenance, they're just holding up a tablet at an engine or something like that. That seems like it wouldn't scale that well to me. Maybe I'm off base there. But is this the opportunity where XR hardware really makes its way into the workforce? 
 

Ian Hughes   00:32:44

I think it should be because the -- and one of the things in the consumer space is that we don't -- we wear various glasses and don't wear glasses and then you get a headset and it fits or it doesn't fit. And then there's a kind of friction to why am I putting this on what what's the point? Can I -- what can I do with all that stuff? But in an industrial context, I mean, putting stuff on like goggles to protect your eyes and a hard hat to protect your head, adding extra technology to that to give you extra information, where it might be a little heads-up display, it might be a mic to talk to your AI assistant that's asking details of all this stuff in the digital twin that we were talking about. That's, again, quite obvious. And it's taken a little while some industries, the whole thing about giving people a piece of tech rather than a clipboard. So giving them a tablet, the tablet and the clipboard are very similar, aren't they? So it kind of -- and then now you've got the tablet to say, here's your work or to tick these things off. Oh, look, the tablet has got a camera on it, it can take a picture of what's going on. The camera can connect to some remote support and somebody can talk you through it. Oh, the camera has got a LiDAR on it. It can work out what the depth of something is, and it can actually point to the thing that you're working with. It's all the mechanics of using those AR approaches on the shop floor are available to those tablets. And then it becomes, well, if we've got a tablet doing that, and I'm now kind of -- I've got one hand on the tablet. I would be really happy if I had both hands free to make fix. -- put these goggles on or put this headset on. It kind of again, becomes an obvious thing. And I've often said to companies, don't wait for the headsets because you can do this stuff with the stuff that exists now and you're not then throwing it away and starting again because it's the same XR that you're going to be dealing with when the headset. 
 

Neil Barbour   00:34:57

Right. Those platforms will travel to the new hardware. 
 

Ian Hughes   00:35:00

Absolutely. Absolutely. It's the same. You've got a digital twin, you've got data, you've got everything. It will be represented there. And there are plenty of unifying data formats and things that are happening along those lines. And that's kind of the day-to-day operational stuff. But then there's the -- it's probably consider it more of a VR approach to training in a place that doesn't exist yet. So to sit in that simulation or standard simulation at your workstation or run the inspection that you were going to do in a VR headset, but you're doing that specifically to say, right, I'm visiting the factory now. So it's like getting in your car and go to the factory, put on the headset, put in the factory. So that becomes less of a barrier, I would suggest, because that's how else are you going to see the factory. You could just sit and look at the floor plan. You could look at some static renders of what it might look like and where the fire exits are or you can be there. 
 

Neil Barbour   00:36:04

You could drive around it in like a Doom-like presentation or something. 
 

Ian Hughes   00:36:09

And there are lots of ways to make it actually interesting as well. So they try and keep it all on the download, but you could make it actually really entertaining. And that's an interesting twist on training, isn't it? Because you do want it to be enjoyable so that your brain takes it in. You don't take in boring things. 
 

Neil Barbour   00:36:32

I suppose the RDS there is probably the cost, but there are -- we've seen solutions, think of like -- I think it's Arbor XR that can sort of do distributed rendering in a way that makes it more palatable from a cost perspective to serve a lot of different units. All right. I think we'll wrap up our conversation on digital twins. Thanks so much, Ian, for sharing your insight and some of the survey data on this emerging vertical. 
 

Ian Hughes   00:36:56

Thank you very much, and it's nice to have been interviewed by you, normally off interviewing many of the other people in the industry, but it's always a pleasure to share where we are with this stuff. And this is data and dimensions, insight across every reality. 
 

Neil Barbour   00:37:13

See you next time.