DUG Technology Ltd (DUG) Earnings Call Transcript & Summary

I might just start by welcoming investors, prospective investors and research analysts to today's introduction to DUG webinar and investor presentation. The objective of today's session is to help new investors who are not familiar with the story to get up to speed with the investment proposition that DUG has to offer. Today, Founder and Managing Director, Matt Lamont, will host the session and we will aim to cover an overview of the DUG business, the competitive landscape that DUG operates in and an explanation of DUG's technology. The format today will be mainly Q&A. I'll start by asking Matt a series of questions and towards the end, you will have an opportunity to ask questions, too. [Operator Instructions]. Today, we've launched a presentation with the ASX, and the presentation will be recorded and posted on DUG's website. One word of caution given the time of the year, we will be unable to discuss the financial performance of the business for the year just gone due to the proximity of today's webinar with the financial results due to be announced on the 22nd of August. Now we've got those formalities underway, perhaps we can start with the first question. Matt, could you give us -- start by giving us an overview of DUG's products and services.

Thanks, Steve. It's a pleasure to be with you all here today. 85% -- 84% of our revenue comes from services, and that's our traditional business, where we take seismic data for oil companies and we process it and we image it and produce an interpretable product for them that they use in exploration and in production. Part of that services is the multiparameter FWI imaging. 11% of our revenue comes from software. We love the software business, and we're always looking at ways to expand it. The software spans from what we do in services right through to how the software an oil company will use to interpret our products. And it's in 35 countries, and it's a great business. And then there's high-performance computing, which is about 5% of our revenue. And that's just where we let clients come on and use our compute. And they come on and they use our compute with their own software or with a third-party software or they can come on and use our software as well. So it's -- and they can -- we have competitors come on and use our software and hardware. So that's how we capture the small end of the market where it's not so financial for us. And places where we don't want to work like maybe the stands. We have a really good client in Pakistan doing a lot of work in the stands through to clients, not competitors and everything. And we have another competitor, sort of a competitor who does a lot of wind farm work using our software and hardware.

Okay. Thanks, Matt. Can you explain DUG's value proposition to customers?

So the main -- 84% of our revenue from services. And we provide oil and gas customers with data analysis, which helps them decide where to drill exploration wells and how to bring a field online and how to produce a field effectively and efficiently. And our software is a part of that process as well. It helps them interpret their products as well as interpret -- as well as process their seismic data.

And can you give us an idea of a typical customer journey taking perhaps an oil and gas customer coming onboard?

So it's sort of different depending on whether it's a small customer or a large customer. So a large -- take an, Exxon, you spend years going through a number of technical -- jumping a number of technical hurdles where they will test your software in great detail and piece by piece. And then when you get over the mark there, then you start to get projects which are in noncore assets on simpler projects in noncore assets. And then you'll work your way up with more difficult projects and work your way into their more core assets. And you work your way up the value chain as you go along in project size and project complexity and so forth. So that's how it works with the major. And we're still on that journey of working our way up that value chain with the majors. With the more smaller companies, you tend to do like a small test project, which you may or may not get paid for, tick that box in a way you go getting the tenders. And that's how it works over time and then they get to lean on you and they really -- they need you and they appreciate you. And with the best client relationships where they like Apache used to say, well, where their technology are what we would get all their work. So that's when you get these really lovely, really tight bonds between us and the clients.

And can you give us an idea as to the typical cost of a seismic program relative to the cost of drilling a well?

So the -- that used to be a really simple answer to that question because the seismic program would cost $30 million to $50 million, for example, and drilling a well would cost $100 million plus -- $100 million up to $500 million or even more. And so we used to say that seismic processing was 10% of the cost of seismic acquisition. And seismic acquisition was -- it could be 10% of the cost of drilling a well. And that's always good to be on the low end of that. So when hard times hit, they don't want -- oil companies won't want to drill wells because that's a lot of money. They would rather do the seismic program. But with the new technology and with the new size of surveys or detailed surveys going on in the Middle East, that's been changed a little bit. So now companies are spending Aramco, Kuwait Oil Company, ADNOC, all spending over USD 1 billion on seismic surveys. And then we're visiting Shell the other day and because of new technology in the multiparameter FWI and elastic FWI spaces, they're now putting 50% of the acquisition cost aside for processing for the new technologies. So things are ramping up in a good way for us.

Okay. And have you done any work on the total addressable market that DUG is playing in?

Not recently. But if you go back 5 years or 6 years ago, we had a meeting with Mackenzie. And they told us at that point in time, the total addressable market for seismic processing and imaging was about USD 1 billion a year. We think it's larger now because of there's a huge surveys going on in the Middle East. And because of all the activity, we think it's probably larger than USD 1 billion now a year. Likewise, for software, it surprised us at the time that the software market was the same size as the services market. And we think that -- but that's probably be growing a little as well in that time.

Okay. You mentioned Abu Dhabi. But is it worth just touching on briefly where you do business, where you have offices and then how the new office in Abu Dhabi is going?

Yes. So traditional offices are Malaysia KL, Houston, and London as well as Perth. We now have office space in Abu Dhabi, and we've just kicked off a fit-out process. We have 16 staff that have signed up, really high-level staff. We're very happy with the work that HR have done and how we've networked and we've chased. Those guys, some have started, some are yet to start in the next few weeks. They're going to Malaysia and Houston for training and then back to Abu Dhabi. And the order book in Abu Dhabi is crazy big. So we've won one job so far with Saudi Aramco, and we're just waiting and watching. The business looks tremendous over there and the reception has been as good as we possibly could have hoped for.

And to clarify, that would have -- I think you mean pipeline rather than one order? Yes.

Not one order, sorry. Pipeline, yes.

Yes. Okay. Excellent. Could you give us the names of some of your larger customers perhaps split national oil companies, the majors and some of the smaller ones. Just so people get an idea of who.

It's easier -- it would be -- if we wanted to cover, which we don't. We just want to be just one of examples, but we work for most oil companies. So we work for Exxon. It's a really good client of ours. Chevron is a really good client of ours. We do production work, 4D work for Chevron. Shell is a good client of ours. Aramco, I just mentioned we've won a project, and we're really looking at developing that relationship and the really taken with our technology, which we're super happy with. We get more than 50% of PETRONAS work. So we've got the majority of the market up in Southeast Asia. We do all work for Santos, and we do some work for Woodside. As I said, it's -- we work for everybody that you could probably think of to be perfectly honest. Total is a great client. E&I is a great client. MX, Petrobras and so forth.

Okay. Sounds like there's a good diversified client base. Can I ask how you contract with those customers? Is it project-based? You signed longer-term contracts or something else?

So with software, that's simple. It's recurring revenue. It's an annual lease and so it comes up and they resign every year. It's very sticky because they've got tens or hundreds of man years invested in their data in our system. So it's hard to shift. We think of services as semi-recurring because although they're one-off projects, generally, they're not always with our best client which is signed, we get all their work where there's a work program for the next 5 years. But that's not normal. Normally, it is project by project, it will come under a master services agreement. But it's not really project by project because we're talking to them. We know what their pipeline looks like. They will be -- we will -- with a major client, we will have 2 or 3 projects at any one time. During bad times, that stream of projects will come down to 1 or 2 projects in good times, it will grow fatter. So we think of it as a semi-recurring business, the services, not a project-by-project base. That's not how we model it up. For a small oil company, it is project by project because they just don't have that stream of projects but not for our good clients.

We've had a question come in via the Q&A function, and I might as well ask it now because we'll just answer that question. If you're working for all of the big oil companies already today, where does the incremental growth come from?

By working our way up that food chain. So if you look at what we're doing for Exxon, for example, we're doing their -- we're doing core work now, but we're doing the simple end of the core work and we're still building our bonafidees to do more core work. The -- and it's likewise. We -- for Chevron, we're doing some really key production work for Chevron in Perth, but we're not doing much work elsewhere in the world. We're doing a little bit of land work in the States, but we're not doing much work for Chevron elsewhere. So you got to build your network within Chevron. Some clients like Exxon are very global we face and they do everything out of Houston, some companies like Chevron a flight distributed. So it's about building up the amount of work we get out of these clients. Petrobras, we sell software to, but we don't do service work because we're not set up yet, but we're on the brink of being set up. Middle East, we're not doing work for yet, and yet they've got an enormous amount of work come out. But we do a little bit of work for all of them. So it's about building our the amount of money we're getting from each client. We've done the hard work, which is getting the first client, getting the first project.

Okay. And do these oil majors do any of their seismic processing in-house or they outsource it mostly?

It depends on the client. Shell will do quite a bit of their own work in-house. Others like Exxon will do a little, but it will be all the testing stuff or their production work will come out. And then other companies like a Chevron will do to basically do it all outsourced. They're just doing R&D bits and bobs internally so they understand what technology is and what's going on. So it varies depending on the client, but all of them give out substantial work.

Okay. Perhaps changing tech, you've invested considerable dollars into some new compute, some Intel machines installed in February and the AMD ones more recently. Do you want to give us an update as to how those new machines are going in terms of deployment? And then actually moving quite [indiscernible]?

So the majority of the Genoa's and all the sapphire rapids are in full production. They are beautiful machines. They are really helping with the higher-end projects where we need that more RAM. We need more RAM than what the KNLs have. So their performance has been extraordinary. We're getting about 6x to 7x the KNL out of our sapphire rapids and up to 10x out of a Genoa. So -- and our R&D teams are still optimizing the codes for those. So we'll get a little bit more out of them yet. But yes, they're beautiful. There's a few genoas that are not quite in production yet that we've only received recently.

Okay. And after all of those MD genoa machines are installed, how much capacity is that going to give you? And I'm asking that question relative to the work that you expect to win over the next 12 months do you have enough capacity?

Yes, at the moment, if things turn out the way we think they will, not in an optimistic way, just in a business as normal way just growing the company in a fairly normal way. We will have enough compute, I believe, to get us through the next 12 months. Now what can offset that are all good things, right? If we win a great deal of work, then we're going to need more compute. Well, that's a good story. If we win a lot of NPF WA, then we're going to need a lot more compute as well. Well, that's a great story. So -- but if we keep growing the way we expect to grow and nothing comes -- nothing is extraordinary, nothing is out of the normal, then we'll have enough compute, we believe.

Okay. Thank you. Can I ask in relation to the compute that you've installed. Can you explain DUG Cool to us and what makes it special?

So DUG Cool is how we house our computers and what our computer rooms look like. So that's our immersion cooling technology. It saves us 50%, 51% of the power bill, which is why we originally did it. We can build data centers really cheaply, and we can house our computers to really run our computers really inexpensively. As we announced before, we are looking at channel partner -- channel partner, channel partners to commercialize that technology. We've been running it at scale for 10 years, which is well ahead of the market. And yes, that's DUG Cool. So we're pretty excited by what's going on there.

Okay. And while we turn briefly to DUG Nomad and the opportunity that presents.

So DUG Nomad we're treating differently. DUG Cool, we're looking for channel partners, someone to take that on board. With DUG Nomad, we've formed a new business unit, and we've hired a new guy to run that business unit. And we think that can be a huge business. And so that's where we're selling a containerized compute system where it has DUG Cool in the container and the container might include cooling and tanks or just be tanks. And we've got a lot of leads and a lot of money, that would be a terrific business. And that's all over the world. We've got serious leads in Saudi, serious leads in India, serious leads in the states.

Okay. Can I ask -- we've talked quite a bit about oil and gas. What other industries do you currently service? And are there any particular sectors you're looking to target for either the services or software part of the business.

So some of the really successful areas that we've gone into outside of oil and gas is [Australia] still love us. They do their modeling of their new hole designs. Bioinformatics, most of the bioinformatic institutes run on our compute, a lot of universities around Australia run on our compute, such as Monash, Burton, UWA have done work, ANU run on our machines. So a lot of that, a lot of exponent modeling is around astrophysics we talked about is ramping up a little bit. So we talked about all those guys before. The problem that we've discovered is that they don't spend a lot of money on compute in the way we're used to people spending money on computing in the oil and gas space. But it is building, I think, although it doesn't look like it. And we're getting really great traction with both the state government and the federal government about how they fund HBC in Australia. Where we're chasing is DUG, as I said, we've got great relationships now that we had EDUC come and visit, which is great. We've had all the government ministers in the state visit great support. And on top of that, we're chasing mining and defense. So we know that it's not gone as well anywhere near as well as what we thought it would or what we hoped, but we haven't given up. We're restructuring it a little bit, and we we've refined who we're chasing and how we're going about it.

Okay. I think perhaps enough on the business just for a moment, I'm sure we'll come back and ask a few more questions. Just turning to the competitive landscape. Who are your major competitors? And if you can give us an idea of their relative size and scale.

So Viridian, who used to be called CGG. I still tend to call them CGG is a formidable geoscience company. They're a very good geoscience company. They have a lot of issues on the business side, a lot of debt and other things, but they're a very good geoscience company. And I think they put out recently that they have 40% of the market. SLB, really good geoscience company and probably have similar amount of the market, maybe a bit more. TGS, we would then come in as a third, we believe, in proprietary processing. TGS do a lot of their own processing for their multi client library. So they would be a significant -- they've got significant processing but of their own data. We partner with TGS, so we're doing a big processing project for TGS in Malaysia, and we've done other processing projects for them from time to time. And ShearWater are a big acquisition, the biggest seismic acquisition company, and they're trying to get into processing, but they're not so good. So that's what the landscape looks like from our perspective. Yes.

And I think I've seen in 1 or 2 of the research reports estimates that your market share is somewhere in the order of 5% to 6%. Firstly, is that a reasonable estimate? And then what do you think is a reasonable share for you to target over a 5- or 10-year period?

So I think that's as accurate as we're aware of in terms of what our market share is. We've been leading -- we believe we're leading technically with our multiparameter FWI. A lot of people believe we're leading technically. But there's a lot of momentum and a lot of belief in these larger companies, but we are gaining market share and we're gaining more rapidly year after year as you get more belief in you. So we believe we can grow -- certainly have been growing relative to our competitors and I believe we can grow more rapidly relative to them. So what can we get to? I think we can certainly get to double digits. But then in markets where we've been really successful and been in there a long time, we've got more than 50% of the market share. So I believe we can grow significantly. So in Malaysia, we're more than 50% of the market. But I certainly think we can get to double-digit share of that market and beyond.

Okay. Thank you. And Viridian and SLB, how do they compete? What are the services do they offer?And what is their competitive advantage? Do they bundle? Do they -- how do they compete against your MP-FWI offering?

So they're quite different, each of them. So Viridian, CGG. They've been the premier processing and imaging group for quite a long time now for probably 10, 15 years, coming off the edge of that now, certainly, but they haven't been. And so they're trading on the back of that belief and you don't get in trouble for buying IBM. They've sold their software. So the cause of their death, they've sold their crown jewel. And they still have a seismic acquisition equipment business, but that has no add-on benefits for their Geoscience business. And so we're head-to-head against them time and time again. They will see a lot of work we don't see yet, but we're building. We see more and more of that work day by day. Their reputation and SLBs reputation in the Middle East, we've discovered is not good, a bit arrogant and not providing what people want, technically. The new technology is not what the people want. SLB is a completely different [cuttlefish]. SLB, again, a very good company, a Schlumberger is a crack company, really well managed. But they bundle. They bundle all their services from their well services and everything, and they will try to just come into a company and take everything. And that does work to some extent with some companies for sure. And so we lose because of that bundling at times. Projects that we know that they want to give to us, their technical folks want to give to us, their procurement want to give to us and suddenly we lose them just because of some bundled deal. And because they're spending so much money because drilling wells and well testing is so there's a lot more money than the seismic. We can lose out that way. But it's nothing -- there's nothing really bad there.

Okay. And one of the questions that I've seen speculated about in a research analyst report or two, do Viridian and SLB utilize their own storage and compute capacity? Or do they go to cloud-based solutions?

So CGG utilized all their own compute. And indeed, they've copied us again. They copied us on a lot of stuff. They copied us on providing cloud compute now to the market. So -- but that's very recently, they've done that. So CG provide all their own. SLB do use some cloud compute, but it's to back up their software. So you can -- whereas -- you'll go on -- you'll go to [indiscernible] if you want to use our software, you can come and use our hardware as well. You can't use the hardware without using the software but there's that bundling where we'll give you a software hardware combo and it will work really easily and really nicely. Well, Schlumberger do that, but they use that they use a cloud compute. I think it's Google, they use for doing that back-end cloud. So if you're using their software, you can run it on Google Cloud very easily. But as far as I'm aware, the vast majority of their process in imaging is still in-house compute. And it just has to be because of the quality of the compute you put in yourself and the cost base that you can put it in for.

Okay. Thank you. I think we're making some reasonable progress. We might turn to an explanation of DUG's technology. This is a complex area. Matt, I'll ask you for my sake as well as some of the investors to go slowly through this, and they might even ask some questions along the way. But could you start with what seismic processing imaging is and why it's important to the oil and gas exploration sector.

So there's 2 types of data, which is critical for oil and gas. The first is seismic, and the second is well data. Seismic is your broad field data. So we cover a very small seismic survey might be 100 square kilometers and a large seismic survey might be 50,000 square kilometers. So it's your broad expanse of data that gives you a look over a big, broad area of the subsurface. Whereas well data gives you a look at a single point on the surface of the earth and both of them are extremely important when you work for 2 of them together if you're an oil company. So all size, all oil and gas exploration and all oil and gas production will use seismic data and seismic data is processed and then it's reprocessed when you learn stuff and or bring out new technology. So it is absolutely critical to oil and gas.

Okay. Can we then ask what is multiparameter full waveform inversion imaging? And how is it different from conventional seismic analysis?

So in conventional seismic analysis, which is old, right, and has had million spent on R&D on it, and it's very -- competition is very good over 60, 70, 80 years now. It takes 6 to 18 months. Sometimes we get not often, sometimes longer, depending on the complexity of the project and the size of the project to do a conventional seismic analysis. And with that, you were using just the primary events. So if I stand, I'd like to use the story of the Canyon. So if I'm standing at the front of the opening of a canyon and I go, Ku-wee, I will get that first Ku-wee back to me. And then I'll get Ku-wee, Ku-wee, Ku-wee and then I'll get this whole jumble of other Ku-wee happening at the end. Over traditional conventional processing and imaging, I will just use that first Ku-wee that comes back. That means that everything else is noise and noise needs to be removed from the seismic data. So I then spend a lot of time and a lot of different algorithms removing different seismic condition different seismic noise trains and conditioning the seismic data ready for imaging. And then I'll go into imaging where I make up a velocity model in a series of standard steps, which take a long time as well. And I end up with a beautiful product, to be honest, it's really nice. But when I do multiparameter FWI, I can get out all the same outputs because the output of conventional, there's probably 6, 8 standard outputs of the conventional processing flow, all of which are necessary and important to exploration and production of oil and gas. With multiparameter FWI, I can put out all those same outputs. So multiparameter FWI is a complete replacement for conventional processing and imaging, but now what was noise is now signal. All those multiples, all of the codes and all of these other pieces that make up a seismic wave field and now signal not noise. And so we've got a lot more signal. It's just one big optimization on a massive compute. So we don't have all the decision making. We don't have a number of assumptions and approximations and workarounds and not -- we don't need them anymore. And so we've got much better physics. We've got a bigger data set because what was noise is now signal and therefore, we can get better results. And because there's much less decision-making and much less man time it's just a big computer, we could turn the projects around really quickly if we have a big enough computer.

And I think the answer was probably in what you've said, but could you just clarify the difference between single parameter and multiple -- multiparameter FWI?

So in multiparameter FWI, in the kernel of the optimization, right, that's -- this is -- and you can think of it in terms of AI, you can think of it like that. It's a physics-based AI, if you like. But in the kernel there, we are teasing about -- teasing apart different pieces of the data. So we're teasing bits of the data that will tell us about velocity, and we're pulling them away from the bits of the data that will tell us about reflections and amplitudes so forth. And we're teasing that apart from bits of the data that will tell us about absorption and we're pulling them apart in a really sophisticated manner in the heart of the optimization, which is the only place they can be pulled apart after accurately. Conventional processing does this over a whole journey of 12, 18 months by hand, whereas we can do it in the heart of the optimization. Once the data has been munged together and you've got your output, you can't do it anymore because it's -- your data has all been mixed up. So with single parameter FWI, they don't pull those things apart. They're not doing that really hard work in the middle of the optimization. They're just munging it all together to get out of pseudo impedance volume. We are the people who invented that, by the way, and brought that to market like 8 years ago. And then from there, you can then derive a pseudo reflectivity body, but neither of them are accurate. They're not accurate. So they're an add-on to a conventional workflow because they don't give you all the outputs that you need and they're not accurate. But they are quick and they are good, but they're not a replacement for traditional processing and imaging.

Okay. I think you mentioned that this recent technological advance is relatively nascent. It's been around, I think, 2 years. Can we ask what were you doing before? How do you grow revenue and market share as rapidly as you had? And then what's the transition like in terms of moving your revenue from what you were doing previously to MP-FWI imaging?

So previous -- before, we always had a technological advantage. We were always the first movers. We'd always think up things. So we were the first guys, as I was saying, to bring out FWI imaging, single parameter FWI imaging. We were the first guys that brought out. We were the first guys that brought out basin probabilistic machine learning for lithology and fluid production. We've had a number of technologies all the way along that are leading that got copied by our competitors one way or another. So we always had that advantage. While we've been building up just the general background of our technology offering as we worked our way up the food chain in oil companies and worked our way up to bigger and bigger oil companies as well. We also cottoned on that we needed to get into the tender system. And so we needed to do all the tests, and we're the third player in the world. And so you're just getting yourself under the tender list, was -- would then give you the opportunity to bid on projects. So we had this technology advantage, and we try to leverage that to get ourselves under the tender this, and that worked well over time, but always took time. But really, to be honest, the big advantage over most of our journey is just service. It's just listening to clients and what they're really wanting and providing better service than what our competitors are providing.

And a question that I've had a few times come over the Q&A session. Can you clarify the difference between multiparameter FWI and Elastic FWI?

Yes. So Elastic FWI, it's a multi -- so multi, this goes back to different wave types, okay? If you look at body waves, so that's waves traveling through a body, whether it's a rock body or a fluid body, you get 2 different wave types you can compression waves or P waves, and you get shear waves through rock, not through fluid. And then if you look at surface waves because I'm sure this will come up to people's mind, then you get rally waves, which are surface waves that look like sheaves, which is your waves at the beach. Okay? So they're different. We're not talking about rally wave. We're talking about -- not talking about surface wave. We're talking about body wave. So body waves, you compression waves and you get shear waves. Conventional seismic, 99% of seismic is done using compression data, so P data and that's it. There is -- we use shear data, but we use it in a sort of underhanded way, but we do that by compression. We only -- we only measure compression, not most of the time. We only record compression waves, and we only put compression waves into the earth. And we use this little bit of change that happens to utilize -- some get some fluid effects out using shear. When we use Elastic, however, we then use fully used shear data and fully use compression data in the algorithms themselves. And that gives you elastic and you can have elastic FWI. Now you've got elastic FWI -- we were first my miles, bringing our multiparameter FWI. We're still a couple of years ahead of our competitors from what they've owned up to with clients asking them. But they weren't standing still. They were working on Elastic FWI. So they are working on enhancing a conventional workflow, whereas we are working on replacing a conventional workflow. The trick, of course, is to be able to do both, which is what we can do. So we now have Elastic as well and we have both single parameter elastic. We also have multiparameter elastic. Multiparameter elastic is not in the marketplace, it will be in the marketplace in the next few weeks before our big conference in America whereas elastic single parameter is, but our competitors have had it for a year, a year or so now and have done well with it.

A question around...

I hope that adds a little bit of clarity.

I think I might need to replay this part of the video again, but I think it's helpful. A question around the sustainability of the technological advantage. It sounds like different competitors are seeking to do different things. How do you see the progress that the competitors are making with Elastic FWI as how does that compete with yours and/or do they also need to develop their own version of MP-FWI in your view?

Oh, they got to have MP-FWI and they know it. So they've been -- we -- if anything about national oil companies is that they will do open technology sessions with all the vendors. So we've sat in one recently and Viridian, CGG and some of them were asked directly, when are you going to have MP-FWI? You have to have it, when are you going to have it. And that's how we learn. When -- and CGG said we should have it in 18 months. So which is quite amazing, it means we're 3.5 years ahead. Indeed, if they have it in 18 months and Schlumberger said it's still a couple of years away. So they know they have to have a multiparameter FWI. The whole world knows that is the [indiscernible] it is the ultimate solution. Our Elastic is now we've got -- now it's in production. And now it needs to get tested, and we need to test it, and we'll learn stuff along the way. So we will be trailing them for a little while because they've been learning and running it for a year now. But having said that, for 2 years or 2.5 years, we've had multiparameter out. And our teams have been working tirelessly on it. So we've tweaked it. We've tested it. We've rewritten bits of it. We've learned enormously from doing projects all around the world. So we're not standing still.

Can I ask what role does DUG's global network of supercomputers play in your service delivery?

They're imperative. I think for those who have been following us, you will see that we ran out of compute, and we paid a fortune to get it from the cloud service providers. And then what you can buy is not the highest quality either, so you can't run everything you want to run. When you build it for your own work, you shape the way you build the network, the way you might tweak the network software even because you know how your data flows across the machines. You know the volume of data you need to get out back across the machines. You know how the machines need to talk to one another and you build your machine to suit exactly what we're doing. And that's at a really high level and you just don't get that from bad vendors. It's very difficult to get.

And do you have an estimate as to what you've spent on your super compute capacity and storage to date?

Over the years, I think we've spent about $80 million in our 21-year history.

And to clarify, that's USD 80 million, I'm assuming?

Yes. We are all U.S. dollar company. Everything in U.S. dollars.

Okay. Thank you. The preference for owning your own compute. I think you've touched on this a number of times through the presentation, but it might be worth just putting even one answer. Your preference for running your own computing storage rather than using cloud-based flexibility optimization. What are the other reasons? What else do you get out of your own?

Well, when our client like PGS, they don't exist anymore, they've got which tells our story. Went to all cloud services. We threw a party. You know as soon as someone goes to cloud, you've got it right, because how to compute will be 1/4 of the price or even the best deals you can get. It will be as reliable and it will be more customized for exactly what we do and within our control. And there's -- the thing to keep in mind is when you're using massive amounts of compute you don't get flexibility from big cloud providers because they don't have that much compute sitting around. So you can't sort of ramp up and ramp down the way people imagine we can. You might be able to if you're using half a dozen machines but you certainly can when you're using thousands and thousands of machines. So if a competitor goes to cloud, and then we throw a party, it's great news.

And do you have a target utilization rate for your own compute?

Yes, although you've got, say, 100 projects running at any one time, you're still -- I don't milled out to give you equal compute load at any one time. So there are peaks and troughs in the compute load. And so you'd like to have on average around 80%, 85% compute load and the rest has headroom because you'll have peaks where you're running at 100%. And you know that if you're over 80, 85, you've got -- your projects are being slowed down.

We've got to the end of the questions that we've written down here. Can I invite any of the research analysts perhaps that cover DUG to raise their hands and you could ask questions directly over the Zoom call. And I would ask if you could keep those questions suitable for a DUG one on one session rather than perhaps something that's a little bit too technical. I see Allan Franklin from Canaccord Genuity, if you could ask your question, please, Allan.

Yes, sure. A couple on -- perhaps just on the technology change. To what extent has that changed or could change the way that you're charging for your services?

Yes, I think that as MP-FWI comes out and it is gripped and believed and we've got clients already in this position, and they see the results getting better, then we'll get more of a premium for the product. While we're still trying to grab market share with it, then we're still competing against traditional pricing models for projects and people -- it takes time to change that because people have budgets in place years -- 1 or 2 years in advance. And so they need to change the budget if they're going to pay more for a project. But we think as is appreciated as a premium project, we are getting more money for these projects, but it's going to -- it's a work in process.

And then perhaps just continuing on the MP-FWI conversation, just to what extent is there a lot of complexity across doing it for land based as opposed to shallow water as opposed to deepwater? And do you think you have the sort of sold for each of those?

Yes, we were in all of those. We've got that beautiful land result we've put out, which got enormous attention. So we've worked -- so that land one is probably a year old now. So we've done multiple land projects. Land are a lot more difficult. It doesn't matter what size or it doesn't matter whether we're doing conventional seismic processing or single parameter FWI or multi-primer FWI, land is a lot more difficult. You've got hills and mountains and sand dunes and you've got a very changing shallow geology, which makes it very difficult. So it is more difficult. It will take time to come to groups with that. But yes, we've been working on all 3 of those for quite a while now.

And then just perhaps last one, please. If you could just frame to what extent your workloads have been in existing mature basins as opposed to exploration style basins?

Yes. Well, most of our work would be in existing basins. In the past, we've done a lot of work in Africa and real frontier areas. But to be honest, now we do work off of Mexico or more work off Brazil. We think we'll be getting a lot of work in India. India will be quite a bit of frontier work. But yes, probably -- look, I'd have to look, but probably 80% of our work is in more traditional basins. I mean there is a saying in the oil industry, if you want to find oil, go to where oil has been found, go to the oily basins. Don't try the frontier areas unless you're really brave.

Thanks, Alan. Jack Daley from Shaw and Partners. You could go ahead and ask your question, please.

You kind of at the start, just talked about the pipeline being quite large. Just how could you frame around the split across MP-FWI than just a normal FWI? And I guess, elastic that you recently brought on and how that kind of contributes to your pipeline?

We don't have I'll answer the easy bit first, Jack. We don't have any elastic projects because we're just in the -- which it's now available for people to buy, but you start off by doing test project for people, and that's where we're at, at the moment. So no elastic revenue yet. A lot of the pipeline is in the Middle East, where it's huge land projects. And so again, we're testing MP-FWI there. We do have MP-FWI projects there. We have other clients elsewhere, Gulf of Mexico and other places where it's just MP-FWI now. But as a percentage of our revenue, it's probably not that high just yet. It would -- maybe it's 20%, 25%. And -- but it's -- it will go up, but conventional is going up as well. Single parameter FWI, that wouldn't be much of our -- we don't even separate it out from conventional processing imaging to be honest. It's not much of a thing.

And is there any way you could kind of frame around the materiality of the pipeline compared to what it might have looked like in the past?

The pipeline compared with what it's been in the past is very large. I probably don't want to say too much more than that until we get out our financials and stuff. But I haven't been saying all long, it's really big and a lot of the big pipeline stuff is coming out of the Middle East. Look, the market in general is really good around the world.

Thanks, Jack. Caleb from PAC Partners. Could you go ahead with your question, please?

Thanks, Steve. Hey, Matt. So just the PGS and TGS merger, now that -- now it's happened. We're hearing how it's impacted, I guess, the industry environment and competitive dynamics and how that impacts the client can do the panels and stuff.

So it really is not going to impact us so much, Caleb, because PGS used to be a competitor, maybe a decade ago, but they've really dropped out of proprietary processing. They couldn't compete. It would be very rare for us to hit them in the tender to be perfectly honest. And so most of -- a lot of their work, I assume, therefore, was some Norwegian work where they would have got in on the ground for and got sole-source project. But most of it was around multi client libraries and processing the multi client data. So that's not us. We're not in that directly in that market. TGS, we do come across a little more in tenders, but not that much, not the major ones. And as I said before, we actually partner with them. So we do processing work for them. So I don't think that merger is that material to us, to be perfectly honest.

Thanks, Caleb. Julian Mulcahy from Evanson Partners.

Matt, just a couple for me. Firstly, how is the supply chain at the moment, getting computers because over the last year, things have been a little bit delayed. Has it sort of got to a level now where whether if you need more compute, you can get it fairly quickly?

We're working -- we -- internally, we're doing a lot more work continuously right back to pre-COVID where we're testing new hardware all the time. We're talking to vendors. We're optimizing codes. So if we need to buy our hardware in a hurry. We're ready to go. And so we're a lot more prepared internally than we've ever been. That doesn't mean we're buying hardware. It just means we're prepared. In terms of externally, I think that there -- there's funny -- there's some funny holdups that we are working around. If -- and it's around things you wouldn't expect. It's around it does a hold up around PDUs, which are power distribution units. So they're sort of like power boards, right, but at a much bigger level for a computer room. And so around that, we can order some because they're not big ticket items. So we're trying to get a few in our back pocket. And we're also looking at, for PDUs. We're looking at a different vendor. So we need them to build us a PDU. So we've come across, for example, is a Malaysian vendor. So we've ordered 1 of this, and then we'll work through the quality of the durability of it, we'll get the best we need in we'll work with that vendor because we know they can supply them a lot quicker than the people in America that are making our PDUs for us at the moment. So -- and it varies over time, but we think we can get compute reasonably quickly nowadays, not pre-COVID quickly, but much more quickly than we have been out to get it in the last year or 2.

And also the second question, given you've got a fair bit of capacity in the Houston facility in terms of space, is there any sort of thought to either renting it out or playing in using some of the compute for AI-related activities for external parties?

We've had several conversations over AI and putting in a machine for AI. It's -- and we do a lot of AI work -- we do a lot of our own AI work, a lot of our programs have AI, kernels have AI pieces in them. And I -- but if you want to go to the main AI market, that's around NVIDIA, NVIDIA GPUs and specific NVIDIA machines. And bang for buck wise, they're about 5x -- 4 to 5x the cost of our machines that we put in. So if I want to multiply 2 numbers together, if I use an NVIDIA GPU will cost me 4 to 5x more than if I use the machines we've been buying. So it's difficult without client who specifically wants a massive AI load to put in NVIDIA machines because you can't compete in what we do as our main business. But we have been working with Intel, at Intel they've got more AI-specific machines coming out and libraries, and we do a lot of AI work for our external HPC as a service clients but not on a big scale. What we are looking at for the computer room in Houston is finishing the fit out of it so that we're ready for our expansion is because we -- when Elastic takes off, we will need more compute but that's down the track right? And that's a lovely problem, but that's down the track, but we need to be ready with computer room.

Lindsay from Ord, if you could go ahead with your question, please.

So just kind of continuing on from some of the discussions we've had on this call. I suppose when talking about your technology with others in the industry, the kind of basic way this is explained to us is you would use or companies would use -- they kind of view CGG as the gold standard in the industry and would use DUG for some of your lower or kind of mid-tier use cases. Now it sounds like from what you've said...

[indiscernible] to the wrong people, Lindsay.

Okay. So yes. So question one is, it sounds like you would disagree with the whole premise of the question. But the real question I want to ask is even people who have gone down that line of reasoning say something along the lines of, well, the geology is actually much simpler in the Middle East than other parts of the world. And therefore, DUG's entry into the Middle East should really suit them perfectly. So I guess my question is, why do you disagree the premise sounds like yes. Second question is, do you agree with the second part, which is the Middle East is kind of -- that should be your kind of perfect environment?

So I think we compete when given a level playing field against CGG, we compete really well. And there are some corner cases in basins where we haven't done a lot of work, where they will still beat us. But the vast majority of the world, we're right up there. And if anything, we can be better, and we've had lots of feedback where we're better. CGG aren't the IBM and you don't get in trouble for buying IBM, and they are technically very good, and they've been technically very good for 20 years. So you can get that result that you've got there, Lindsay. But I also, we've heard from some of the other analysts here where they've gone out to experts who say that we're leading the field and we're taking over. So now in terms of geology in the Middle East, it has its own challenges. A lot of its challenges are around multiples, and we're very good at multiples. So multiple is where you get the seismic bouncing around in layers. And multiparameter FWI again is really good at multiples. So yes, I'd disagree with the premise. I would say that CGG have been the technically leading company for a long time, but they're not now. CGG would absolutely disagree with that, and you will have experts that disagree with that, but you'll have experts that support us more than CGG. So we're coming. They know we're -- they used to ignore us, but they certainly don't ignore us anymore.

That was brilliant. And then just the second question is just on the pricing and margins for, say, MP-FWI versus some of the traditional workflows. I mean, from everything you've said intuitively, you would think you'd be able to charge an exorbitant premium, given that you can cut down timings by 3 to 10x, let's say, from my understanding is that's not really the case. So could we just talk through pricing for MP-FWI and the margin profile versus your traditional workflows?

So we are 5% to 6% of the market. So we're a small player with leading technology. there's this belief piece that you've got to get out there and you've just got to do a lot of projects. And there's this whole [what] budget to oil companies have to do a project and so forth. And is this whole belief piece the technology people in an oil company are likely to believe in DUG, a lot more than the asset people with oil companies. And so in some -- so the asset guys are the geologists and the people that are using the data and they've been using CGG data for 20 years. And they've seen DUG data and they're okay with it, but they've -- you know what I mean, they don't want to compromise. And whereas the technology people will know that it's not a compromise, but they've got to sell it internally. So there's this whole process going on as we bring out MP-FWI, and we do more and more. And we've got clients now that just say it's fabulous. And we just want it, and then there's other clients who're still questioning it. And of course, the backdrop to all of this is you've got companies like CGG just muddying the waters as much as they can -- and we've seen internal documents where they'll lie outright to clients, but they won't put it in the public arena, but they are doing something that's supposed to only be for the oil company and not shown publicly, then they'll lie outright about MP-FWI and what they've got, what they're doing. I'll say they [indiscernible] CGG if they think the docs not going to be public. So we're in a battle. We're in a fight. Now we're going to win the war, but we're not winning every battle. It takes time to get this through.

Thanks, Lindsay. [indiscernible] Wood from Wilsons Advisory.

I'm just going to ask this [indiscernible] is commute. But just one for me. I guess as you go from this smaller harder task for client, I guess, to the larger...

You're breaking up a little bit.

Sorry, can you hear me now?

I can. Yes.

My question was just as you go from, I guess, smaller harder task to, I guess, the larger, more general task. Just how on a project basis, does that kind of change from a margin perspective?

The -- we do -- when we do small work, we do really high end, more work and more general work easier, harder work. But there's a really simple rule. Bigger projects make more profit. Just simple bigger projects suit us more as well. Because we're so good at compute, we have such a lot of compute. So you really -- it's really simple. You really want to just do bigger projects. And then the secondary factor is you want to do simpler projects. So you want to do bigger projects, simpler projects. And often, as you're growing, you'll get smaller projects that are more complex because they're the ones other people we don't want. But what we're seeing more and more now is bigger projects, and that's what you want. They're more profitable.

And then I guess the second question was just on DUG Nomad. Just given like the pipeline opportunity you're seeing there, I guess, do you have any color on, I guess, the sales cycle and how long it would take, I guess, to fully land a client?

No, because we haven't sold any yet, I thought we would have, but we also haven't really -- we're still working on getting our supply line set up to deliver them and getting that whole business unit up and running. So we're a bit behind where we want it to be, but there's a number of technical challenges we've had to overcome. And now we've just shipped our first one to America to a client for testing. And ask me this time next year, I'll certainly have a lot better answer around Nomad for you.

Okay. Thank you, Matt. We've just gone through the hour. We might leave it there, if that's okay. Thank you, everybody, for your attendance today. You will receive an e-mail after the session with a link to the recording of today's webinar in case you would like to watch listen and/or watch it again. It will also be on DUG's website. Just a quick heads up. DUG's results for the FY '24 year are out on the 22nd of August. We intend to host another webinar at around 10 a.m. first time or mid day Australian Eastern Standard Time on the day. Details will be launched on the ASX. And then lastly, we intend to hold a roadshow in Sydney and Melbourne on the week commencing the month of September. We are being hosted by a number of brokers in group meetings, breakfast to lunch over the 5 days. And we will also be doing one-on-one meetings. Institutional investors that would like to participate in the roadshow are asked to reach out via e-mail either at [email protected] or [email protected]. Thanks very much for your attendance today, and we look forward to seeing you at results.

Thanks, everybody. Bye for now.