Sims Limited ($SGM)

Well, good morning, everyone, and welcome to Nashville. For those of you who are watching us on the video, you'll be pleased to know that we did make it. If you're watching yesterday's video in Houston, I can confirm that it was a 3.5-hour queue at Houston Airport, but we got through it. That was an interesting experience. And here we are at day 2 of the Investor Day looking at Sims Lifecycle Services. We've just been on the site tour. Thank you very much, Isaac. Really appreciate the tour. It was really good to get your insights about how we go from when the material arrives right through to when we finally dispatch it for resale or back to the redeployed unit. Thanks very much for that tour. The agenda for today is I'm going to give just a really quick introduction, I'm going to hand over to Ingrid. Ingrid and Sean will take us through a presentation, then we'll have plenty of time for Q&A. We don't need to rush to an airport like we did yesterday. My first slide is entitled why SLS matters to Sims. I think after you've been on the tour, it's quite self-evident. I just want to make a few high-level comments. This has not been an overnight success. I know from the market's point of view, SLS has grown rapidly really for the first half of '26. And clearly, we put out some guidance for the second half of '26. So it appears like a rapid growth. We've been in this business probably since 2019 when we sold off another part of the business and decided to focus on hyperscalers. And so we've been building up relationships. We've been building up our operational capability. We've been building up our sales capability, our redeployment capability over that time in preparation for this. So I'd like to point, it's not just an overnight success whatsoever. And SLS now is a key part -- clearly is a key part of Sims. It's a key part of our growth. If you look at the chart here, I mean, it puts it quite starkly the EBIT contribution from SLS, how it's grown over the period. And by the time we get through to the forecast year-end, it is going to be a massively significant part of our business and therefore, an area that we've been working on, and we're really pleased that it's now getting the limelight that it deserves. I think you'll see throughout the presentation, it's the skills that we've developed. It's the relationships we've developed with hyperscalers, how we're embedded in their business, the trust that they have with us. This is, I think, the magic sauce in many ways to the SLS business. It's complementary to our metal recycling business. In a sense, they rhyme to create a world without waste to preserve our planet. It's a very logical thing for us to be doing what we're doing at SLS. But in other ways, it's completely independent from the metal cycle or the metal environment as well. So we've got ourselves another significant EBIT stream in the business, complementary but independent to our metals business, which I think creates a better balance sheet. It creates a more reliable cash flow, creates a more reliable EBIT. I really do think it is very, very complementary. The second last point I make on this slide here is capital light. And I don't know for those who remember, when we first talked about the SLS business, we referred to it as a no-regret strategy. It was always going to be a capital-light business. Yes, we'll be investing some money in robotics, as you saw out on the floor today, but it's not significantly large amounts of capital to get that in place. And so I remember saying right at the beginning, if we turned out to be wrong on the SLS opportunity, which we haven't turned out to be wrong. But if we did, we weren't putting a huge amount of shareholders' funds into it. It's a very capital-light strategy, very, very high return on those assets. And what the SLS business, the final point I'll make before I hand over to Ingrid is we really do have a good understanding of the secondary markets, whether that be redeployment back into a data center, which is a skill absolutely unique to SLS. No one else does that or whether it's selling into the secondary market. We really do understand those markets. We understand what they need. They understand our quality, they understand our processes. They understand that when you get a refurbished, I guess, what we're talking particularly we'll focus on today, you're getting one of extremely high quality with SLS. On that note, I'll hand over to Ingrid and then Sean, and I've got a really exciting presentation for us, and then we'll come back for Q&A.

Thank you, Stephen. All right, so it's me. So I'd like to introduce you to my executive team. We're independently managed from Sims. So I report to Stephen, but my exec team is completely independent and independent because we tend to move at a different pace from the other parts of the business just due to the nature of the business we're dealing in. So Sean Magann, my Chief Commercial Officer, who's here; Lynn Jacobs, who used to be my CFO last time we had our Investor Day and is now taken over operations, so Chief Operating Officer. Chris Guarini, who joined us from Metals. He's in a new role, Chief Digital Officer, and that's because we felt we needed to elevate digital. So he's going to be all about digitally transforming the business so we can continue to stay at pace with our clients, very fast moving. So Chris has a load of things on his agenda to go through. And Jim Clark also joined us from Metals, and he's our new CFO. And then Marie Burke, our Chief People Officer. And we are all headquartered in Irvine, California. So we felt it really important for our team to be together so we could move quickly and make decisions and keep strategy moving forward at the pace. And why California? California because that's where most of our clients are, headquartered in the Bay Area, so very much California-based and plus it's excellent. Right, Stephen? So SLS, global leader in circular cloud solutions. Global, we are in 11 countries and where we aren't, we'll use subcontractors. So we use up to -- we're in 60 countries in addition with our subcontractors. So some of our clients need services, somewhere where we're not placed, we'll use subcontractors. Global, because our clients are global. We need to be consistent. The reporting has to be the same. It has to be sustainable. That's why it's really important for our business to be globally present. Circular, we're all about extending the life of electronics as much as possible. So doing the reuse, the reengineering, redeploying and at the very end, recycling. And cloud, we positioned ourselves, as Stephen mentioned, in 2019, we really turned away from electronics recycling, so we could focus more on hyperscalers and taking advantage of the growth in that area. So we've positioned ourselves to be adjacent to the growth. Who are our clients? Well, as you saw out on the floor, hyperscalers make a big part of it. Enterprise clients. So we have banks, insurance companies also have on-prem services, hyperscaler data center type services that they need as well. And OEMs, so the folks that are making the servers and the racks are also our clients. So we have about 150 clients under contract, and Sean will talk a lot more about that when he comes on later. So as I said, we positioned ourselves strategically to take advantage of the data center growth that's going on that we're experiencing today. That was a strategic decision we made in 2019 to really focus on this area. And we operate very close to hyperscalers to take advantage of that growth. We partake in the technology refresh cycles and of course, work in the supply chain. So we provide a lot of parts to go into that circular in the hyperscaler space. Stephen mentioned, it took us a lot of years to build these trusted relationships that we have with these clients. We've been working with them for a long time. This site here, it took us 3 years to get the technology approved, the process approved to get it going. But now it's -- you see it's running very well. We provide all the secure services. That's very important. What you'll see, Isaac mentioned all the cameras we have, security is of the ultimate importance in this area, about data, anything with data, is super important. And what we've proven is that we can move quickly when they need capacity. That's the big part. hyperscalers need capacity. And it normally comes in -- when they decommission, it's big slugs. So what we do is we provide capacity when they need it. We can put a site up in 3 months, which we've done 3 times here. And the high quality. Isaac mentioned how important it is to have quality because we are competing against new material. So manufacturing like process material. It's a manufacturing-like process. So the volumes to growth, what Sean likes to say, we got to stay on the horse and just continue taking part in the pie growth as it grows, we grow with it, grow with our clients. We're adding new services to continue to grow. We're going in new geographic areas like taking part in the pie growth has -- it grows, we grow with it, grow with our clients. We're adding new services to continue to grow. We're going in new geographic areas like Ireland, for instance, that's also to provide capacity to a client that request it. And then also continuing with our core base of the enterprise clients. So the banks, the insurance companies still need our services, and there's still some growth in that area. So you might ask why? Why the hyperscalers outsource? Why do they do it? So they can focus on their core, right? So they can focus on their core innovation instead of taking space and personnel to do what we do. We provide operational efficiency because decommissioning cycles tend to be very lumpy. So they can't handle the big sloth of material and then reduction of material, whereas we can because we can overlap different decommissioning cycles from various clients. And then we just help with economic recovery. We help them do a more efficient operation, get some revenue back and also repurpose where it makes sense. So as you saw on the floor, this is the main process. They decommissioned equipment comes from the hyperscaler sites, comes to us. We harvest what is requested. We assess which way it goes. Does it go for redeployment? So testing and preparation relabeling. Does it go to resale where there's rev share or does it end up being recycled, which is a commodity recovery. And with that, I'll hand it over to Sean.

Thank you, Ingrid. Very nice to see everyone this morning, it's nice to see so many familiar faces and some new faces as well. As Ingrid mentioned, my name is Sean Magann. I'm the Chief Commercial Officer here at SLS. I'm based in the San Francisco Bay Area, as Ingrid mentioned, but I spend an equal amount of time at our Irvine office. I did undergraduate work at UC Berkeley and from there went on to get an MBA. I've been in the industry for about 30 years. I was very young at the time. But I started with a Canadian smelter that would process electronic scrap and import that material into Canada for precious metal recovery. And one of the projects I worked on that I'm quite pleased with was we set up preprocessing sites all over the world, specifically in Southeast Asia and would process the material in country and then ship and would ship the semi-process material to Canada for smelting. And what I learned there is to maintain margins and to really keep customers happy, there's 3 pillars of service that we have to maintain. The first is service itself, right? All customers have SLAs, and it's critical that we meet these SLAs of customers. Isaac does a fantastic job with Crystal here in Nashville, and our customers are very pleased, and that's one of the reasons why we're offered new opportunities is because we consistently outperform on our SLAs. The second is compliance, shipping material cross boundary. It's a multi-jurisdictional process, and there's a lot of different laws that govern the movement of this material. And so it's critical to be compliant. I don't know if there's anyone heard about Super Micro recently. Have you seen the news when you have a chance to read about it, it tells you what happens when you're not compliant with shipping some of these materials internationally. And the third is capacity. As Ingrid mentioned, our business tends to be lumpy. So you have to have the capacity to meet customers' needs. And that was something we employed at my past company, and that's something I'm very pleased that I was able to bring -- that we were able to bring to Sims. So how do we make money? We have 3 basic buckets of money -- of revenue, I should say. It's resell, service fees and commodity. We'll start at the top with resell. Each of these shapes roughly approximate the relative percentage of revenue we get. So resell is the top now. And essentially, what we do simply put, we take whole units such as laptops and servers and desktops as well as components, hard drives and GPUs, CPUs and memory, and we'll process those materials on behalf of our customers. We'll wipe the data. In some instances, we'll load an operating system, do some of the work that you saw out here with relabeling. So we'll process the material and sell that material to our worldwide buyers. Now my team spends quite a bit of time vetting and approving buyers globally to make sure that they're compliant and that we also receive a fair value for the material. In general, there's a 70-30 revenue split between the client and SLS with the majority going to the client. It works really, really well because it incentivizes us to try to find the highest value because we price participate. So again, we spend quite a bit of time finding the best sources globally to take our materials in a compliant way. The second bucket you see there are service fees. So most items that we receive will have a service fee usually per unit. In some cases, we still have service fees per pound, but we'll charge the client per unit per pound to process the material. What's nice is, in some cases, we can double dip with the same unit. So the same unit, we may have a service fee, we may sell that unit and also price participate on the resell. But the processes -- the service processes can include redeployment where we put something back into a data center, drive destruction, which I think you saw a little today, remanufacturing of units and also data center decommissioning. So that's the second bucket there. And then the final bucket is commodity revenue. Again, we process something for something that's not available for redeployment or resell, we'll go ahead and sell that for commodity. And again, we'll enjoy a revenue share with the client. What I really like about this slide is 5 years ago, 6 years ago, the recycling would have been the majority of our revenue. But because we consciously put ourselves in a position to go upstream, we've really balanced our revenue streams where resell and service fees are much, much larger than the commodity side. Why? It's more repeatable. The margins are better, and it's something we can control, can't control commodity prices as we know. So this slide may shock you guys. I want to make sure everyone is sitting down before you look at this slide. It's no big news to anyone here, but data center spending is going up exponentially. A few notes on this slide. You have 2 lines, one is global, one is domestically here to the United States. This only includes hyperscaler spending. Now there's no true one definition of what a hyperscaler is. So we list the hyperscalers that we include on the bottom there. I hope you can read that there. The question I get asked most about when this slide comes out is folks say, well, does this include land acquisition, concrete, external cooling? It does not. This is only for their IT spend. This is how the data centers make their money on processing and doing AI and whatnot. And this spend is for material that we could process. So you see it's a pretty, pretty steep curve there. Now one thing to note, too, is there's a delay between when something is installed and when something comes to us for processing. Typically have been about a 5-year end of life from installation to decommissioning. We've seen those time frames compress. It's on average, 3 years now we're hearing with some of the AI equipment. We actually have someone well placed in the data center industry saying that, that time has actually been compressed down to 2 years. So we're seeing that lifespan shorten. It's interesting to note that the relationship the hyperscalers have with their data center equipment has changed. Before it was an asset you would sweat long term. Now it's almost a consumable, something they use and discard to keep up in this AI arms race. But what I like about this slide and hopefully, what you as investors like is with that lag, let's assume it's a 3-year lag. Some of the equipment you saw like even this rack here was probably put in place in 2021, but we'll assume a 3-year refresh cycle. So that means the material that we're seeing, even though we're having a really, really great year is for material that was probably installed in 2022, 2023. So that was before this exponential spending. So if you put 2 and 2 together, it seems like we have a really, really bright future. I'm not giving guidance, but you just see the amount of volume coming -- going in would be the amount of volume that has to come out. And this is why we're excited about our business and hopefully, why you guys are excited as well. I can answer questions on this later at the end if everyone has any questions on that. So memory, right? Sorry. So memory, there's a lot of talk about DIMMs, right, DIMMs, DIMMs, DIMMs, specifically DDR4 memory, right? The price has gone up exponentially in the last year. And the question we get as a leadership team is, why is it going up? And is it going to last? And hopefully, by going through these 3 bullet points, I can answer that question for you. So what's going on? So number one, DDR4, which is the second most utilized version of memory, DDR4 is being removed, right? It's not being idled. It's not being turned down until prices pick up. It's being permanently removed. And so if you look at the bullet point or the circle at the bottom, it says this is not a pause in production. It's a deliberate strategic exit from DDR4. Samsung, Hynix and Micron all announced last year in summer of 2025 that they would be ceasing production of DDR4. Those 3 companies make up 95% of DDR4 manufacturing. So they've been very public to say we're out of the DDR4 business. The last buys to the hyperscalers is this year. So as we understand it, as of next month, the hyperscalers are done. They cannot buy any more DDR4 on a large scale. So you have one, the supply being constrained. So the next question I get is, well, why? Where is the money going, right? You said a bunch of money being spent with data centers. Where is this money going? Well, it's going to DDR5 and something called HBM, which is a high-bandwidth memory. So the amount of spending going into manufacturing of DDR5 is unprecedented. As you see the chart there, it was $29 billion in 2024, $53 billion in 2025. And since this chart was created 4 weeks ago, the forecast for 2026 has gone up to $71 billion. So money is being spent. It's just not being spent on DDR4. So all the manufacturers have placed their bets on DDR5. Question, why do the hyperscalers want DDR5? Why isn't DDR4 good enough? There's a lot of different attributes on that make DDR5 better than DDR4. But the main 2 attributes are speed. DDR5 is a lot faster than DDR4. And again, in this AI arms race, speed is everything. And it's also more efficient with power. And one of the big constraints of data centers is energy. There's just not enough energy to do what they need to do. And so anything that's faster and better with energy is much, much sought after by the data centers. And so because of this -- so now we know why the hyperscalers want it. The reason the manufacturers want it is because they can make a lot more money. Per unit, it's estimated between 5 and 10x the amount of revenue selling a DDR5 versus a DDR4. Even though short term, the price of DDR4 is going up, the manufacturers know that long term, their future lies with DDR5. And so now we have a -- and so now we're comfortable with that there's a supply constraint with DDR4. How about demand? There is a huge locked-in base for DDR4 demand. So this isn't hyperscalers. This is everyone else with a data center, colocation site, et cetera, that they've got assets that again, that they will normally sweat these assets between 5 to 7 years, and it's all built around the DDR4 platform. DDR5 in general, isn't backwards compatible with DDR4, the exception being the CXL that we saw out on the floor, but that's specifically a hyperscaler technology, and we can talk more about that. But in general, it is not compatible. And so if you're running a system, if you're a bank, an insurance company, again, a smaller tier data center company, you have 2 choices. You can either replace -- if you want to upgrade your system, if you want to add capacity or upgrade your system, you can either replace the whole system, which is built around the DDR4 ecosystem which is expensive or you can pay the high prices on -- the elevated prices for DDR4. Depending on what you read, we expect this demand to last 3 to 4 years. Again, there's a lot of articles about it. You can Google search on your own, so it's not just me, but we see the demand lasting for the next several years where we all know that the supply is going away. One just quick thing to add about the supply. One of the questions I get most often. I think one of you fellows asked me about it last night was how about China? Is China going to fill the gap for DDR4, right? Can they come in and make a lot of inexpensive DDR4 and try to fill that supply gap. In my opinion, based upon what I've read, that's not the case. The Chinese government has been very, very specific telling Chinese manufacturers not to invest in last year's technology that they want the Chinese manufacturers to invest in DDR5. Again, AI is an arms race. It's an international arms race in a sense. And the Chinese government does not want to be one generation behind. So it's very, very unlikely to see China enter into the market in a big way. There may be some odds and sods that may fill a small gap. But in general, the supply gap should last for many, many years. And finally, this is just an overview of the transition of memory technology. So the version before DDR4, you guys want to guess what was called as DDR3. And we've seen this transition before from DDR3 to DDR4. When that happened, there was a spike in prices. So we've seen this before. Every time someone talks about prices going up and when the question comes up, is something a bubble, the answer is always it's different this time. But it is truly different this time because what's happening now, when the transition from DDR3 to DDR4 happened, there wasn't this rush. It was a gradual transition where manufacturers still supplied a DDR3 and the demand for the DDR4 wasn't as great as it is now. For DDR5, 90% of all that capacity I talked about earlier with that spending, that $71 billion this year, 90% of that capacity is already spoken for by the hyperscalers. So you have the supply -- sorry, the demand side being extremely, extremely aggressive. Well, again, on the supply side, it's being turned off instantly. That did not happen before with the transition from DDR3 to DDR4. It was a much, much gradual. Even though we saw a spike, it wasn't as prolonged as I expect it to be now because the fundamentals are quite different. I've given you a lot of information. Happy to take any questions later. I can talk about memory all day long as you probably see. But with that, I'm going to turn it back over to Ingrid. So thank you.

All right. Well, let's pull it all together and walk through what this means to the financial results of the business. The first point that I want to bring you to is that we're introducing a new volume metric. In the past, we were using repurposed units, and that could have been anything. It could have been a laptop. It could have been a memory stick. It could have been a switch. So we're turning away from repurposed units, and we're going to go to gigabytes of memory sold. So the reason being, if you look here, you have 2 sticks of memory. In the past, there would have been 2 repurposed units. But a 16-gigabyte module is not the same as a 32. It's twice the value. And for those of you that are counting the chips, remember, it's mirror image. You've got chips on both sides. So we're dropping the repurposed units going to memory gigabytes sold. And in future slides, I'll explain why this is important for you to follow when you're looking at our financials. Okay. Sorry. We're standardizing the memory because it just makes more sense for our business as we're going forward. And we've spoken to you before about TrendForce. So this will be the page, the landing page that you go to, if you go to trendforce.com. -- and I'd like to draw your attention to the red box there. That's what we use as our benchmark, and it's a DDR4 3200, and that's the speed. So that is a benchmark that we use to help us guide in our pricing. And if you remember, at the half year results, we said that we were about 50% of that price. Well, that was over here in the gray part. And what we've seen starting in December is this huge hike in pricing. And so really now we're at a 25% to 30% of that pricing of the TrendForce pricing of a new gigabyte. So the first half and different -- second half were quite different. So let's pull this all together. There's a lot of information on the slide, and I think you're going to -- have some good information for you. Last week, we posted a trading update of our underlying EBIT for fiscal year '26 to be between $165 million and $185 million. So what we're doing now is providing you more detail, especially around gross margin. Sean mentioned the 3 buckets of revenue. So our 3 buckets of gross margin are resale, service, service fees and the commodity. And certainly, the resale is the largest gross margin bucket. 80% of that is enjoying the tailwinds from memory pricing. 20% is your consumer tech. And that's your probably slower growth, PC sales, laptop sales. There will be some growth, but not to the same extent of 80% coming from the memory. So 20% of the gross margin is consumer tech. So that's about $40 million to $45 million of this $220 million to $230 million. Service fees we'll see some growth, but again, not as much as the tailwind we're seeing from memory pricing. And commodity will be probably like your inflation. You can decide what it is in commodity, but not to the extent that we're seeing in memory. And we're forecasting our memory gigabytes sold for fiscal year '26 between 65 million and 70 million gigabytes. And what we'll have to point out here as well, the first half, the split between first half and second half of the gigabyte memory sold, we saw a higher 60% in the first half, and we're forecasting 40% in the second half. And that's because we're seeing more 16 gigabyte memory modules coming out in the second half as opposed to the first half. Okay. So Ireland, we talked about the geographic expansion, how we're going to continue to grow. We expect to be operational by the 1st of July, so our next fiscal year. We estimate in the first year, we're going to have 4 million gigabytes sold. And as we continue to scale up, it will be 15 million gigabytes sold by fiscal year '29. You can estimate an operating cost of EUR 4 million in '27 and EUR 5 million in '28. Plenty of room for other hyperscalers to come in. This is dedicated to one client. So we go where the capacity is needed, and this was with partnership of our clients. So plenty of room to add others. Sean is going to work really hard at getting other hyperscalers into this site. This is very exciting for us because Ireland is certainly the largest concentration of data centers in Europe, and it will be a great footprint into that market. So we're structured for growth, as I've mentioned before, we have plenty of capacity. We're probably using about 30 million to 40 million -- 30% to 40% capacity here of 60 million memory of gigabytes for the U.S. and we can add additional capacity as needed by changing some of our sites around. So we have plenty of capacity to continue to grow and to take advantage of this attractive market. And we also future-proofed it by adding automation. So the automation will allow us to continue bringing in efficiency and also to dial up and dial down as needed with the decommissioning cycles that come through. And it also will be compatible for DDR5 once that comes out. So in conclusion, the key messages I want to leave you with is that we're already a material earnings driver for Sims, which is quite attractive. We are exposed to structural noncyclical growth. We're embedded with hyperscalers. We're capital-light and a high-return business model. We have multiple diversified revenue streams. We are taking advantage of the attractive structural tailwinds in the memory markets, and we have significant growth runway ahead of us, so we can add capacity as needed and continue to grow. So with that, we'll turn it over for questions.

Lee.

[indiscernible] just on the pricing side, so why do you look at an 8 gigabyte price when you sell 32 gigabytes and 64 gigabytes?

No, so that is a chip. So that's 1 chip.

Okay. And then if we think about the...

So sorry, so you would multiply it out. Okay. So 1 chip, that's new 3200 speed chip, and you would multiply it out by 32 or 16 depending on the module to get the price reference and then take the discount to what we're getting.

If I can add too, I think the confusion might be that what you looked at that says 8, it's G little B, that's gigabit, which is 1 gigabyte. I know it's really confusing, but that chip is only 1 gigabyte. So even though it says 8 G little B, that means 1 gigabyte capital B. So that's why we use -- there's 1 gigabyte as a baseline.

8 bits and 1 byte.

Okay. [indiscernible].

So that's why that makes more sense, right? We're not basing it on a [indiscernible] 8 to 1. It's 1 gigabyte.

Yes, that makes sense. And then if we think about the lags that you've talked to in the past of, call it, a month and there's maybe some flex around that if the customer wants you to hold back and not sell immediately. Most people, I guess, track a new pricing series because that's what we have access to. And it seems the pricing has stabilized over the last few months. So I guess what I'm trying to work out is how much of the spot price that we're seeing out there is in the guidance that you've given for the first quarter. Is there some equivalent price that you can say in that series that's baked into the EBIT guidance? What's your assumption...

I'll handle the guidance question [indiscernible] so we've pretty much assumed that -- as you said, that price has risen to that level and stabilized. That's what we've assumed in the price in our guidance. And we've assumed that we get 25% to 30% of that price. So the lag effect is finished, I think. I mean, as it ramped up dramatically, it was a little bit hard to say exactly where it was. But as it seems to have settled at this higher price, that's the price that we've used and 25% to 30% of that is what we've assumed we will receive for the second hand.

I mean, you obviously have a volume story here as well. But if volumes were to stay the same, you would print a similar number of earnings for the next quarter.

Yes, yes, although bear in mind, as Ingrid pointed out, which is why we've gone to gigabytes sold we are seeing more 16 -- and because we can look into the systems, we are seeing more 16 gigabyte in the second half. I wouldn't read too much into that, though. I mean this is just the particular ones are being decommissioned. But we've said 65 million to 70 million. And if you look at -- we've also said that about 60% of that happened in the first half and 40% of that 65 million to 70 million will be in the second half of gigabytes sold. Interestingly, more repurposed units in the second half, but because of 16 gigabyte, less gigabytes sold. I think as we get used to that as a measure, you'll understand the benefit of us moving to gigabytes sold as a much better indication of future revenues.

And if we were to look on that metric, what would have FY '25 or if you go back, what's the level of growth we're seeing in that?

So in FY '25, interestingly, because it's recycling, the gigabytes sold would have been around about 5% higher, to be frank. But the repurposed units in FY '26 versus FY '25 around 22% higher. And it's all to do with the mix at the time between 16 and 32, we had some more 3 gigabytes in that 3 -- DDR3 in that period as well. So what I would say is we're seeing good growth in volumes. The mix this year has been slightly different to the prior year. But again, I mean, I think the point I really need to make is don't read too much into that. It just depends what particular data center is being decommissioned at that time.

And then on the volume side, like we heard on the tour just about the lumpiness that can be in the business just in terms of deliveries. In terms of the capacity and installed capacity, how do you think about adding additional capacity in a world where utilization is relatively low, but it's very, very lumpy and you obviously don't want to disappoint your customers by able to not take a shipment.

So as we're moving off guidance questions now, I'm going to hand that back to -- but guys, I have a general view of how we -- but I'll hand it back to Sean and Ingrid. I mean, I have a general view around how we manage capacity and we manage it well with a number of customers. But let me hand it back to those guys who actually have to manage the cycle.

Right. So we manage capacity by adding additional shifts and with the automation you saw, so you can dial it up, dial it down. And we have plenty of capacity to continue to grow. Sean is going to fill up our sites.

And then maybe for Sean, do you want to give us an idea of how you're tracking R&D about reusing or recycling HBM and rare earth and some of the other opportunities that you have for the business?

We're lucky enough to where they're a bit separate, both of them. So the rare earths, there's several companies, a lot of these with government grants that are looking to have a better story for some of the rare earth. So we've got NDAs and relationships with 4 of the big companies doing that, both domestically and internationally. On the HBM, it's a little bit trickier, right, because you've got the memory manufacturers and you've also got the NVIDIA, the folks making the GPUs. So we're working with our hyperscaler customers. It's really different now that we have a relationship with hyperscalers because we can see a line of sight of what's coming out. So we're working with the hyperscalers. This is why we have those test racks to figure out the best way to maximize the value for the stuff they get. So our line of sight to answer your question, is working with the hyperscalers. They know what's coming out. We know it's coming out. So we're working with them to try to find the best homes.

And then maybe just a last one. Your view chatting to either your customers or suppliers around the relationship between used and new pricing, like do they think that the 35% holds? Or do I think we go back to the traditional 50%?

Well, yes, we're kind of -- that's why we didn't give out so much guidance at the half year. We're waiting to see where things are moving. So far, it looks as though the pricing is staying the same.

It's stable.

It is stable. So we would expect it to go back to 50%.

Yes. With that stability because you're not having that big price curve, I think both the spot and the contract -- or sorry, the new and the used price are starting to normalize a bit. So I suspect we don't have enough data to prove that, that is what's going to happen...

It saves a bit more time.

But it makes sense if that would happen.

I think it is important to say, though, that we didn't assume that in the guidance for FY '26. We assume the 25% to 30%.

Yes. That was why I asked the question.

A quick question on the competitive environment, particularly with the hyperscalers, where are you winning? Where are you not winning and why?

Well, certainly, for us, what we do, we're the only ones who do the redeployment. So that is something that gives us very good relationships with hyperscalers, shows our ability to meet their needs because in redeployment, they're after the parts, the parts they want to go back into their data centers. So that is a competitive advantage for us because we're the only ones that have that technology. And it took us 3 years to get that approved to get through all the engineering specs, the security specs and that -- so it's not something that can be easily replicated. I would say. And then on the other side, do you want to add more on that?

Yes, the customers we work with, and it's not just here in Nashville. If you go around, we've got different sites around the country and around the world that are sort of semi-tailored to that particular client. I'm talking thinking of Atlanta with the other hyperscaler. And so with the hyperscalers we work with, we win the majority of the time. There's still a few hyperscalers that we don't work with. And one of the reasons it's not so much the competitive environment. It's more that they're doing it themselves. But I suspect that over time, that can change. Ingrid mentioned earlier in her slides why hyperscalers look to outsource it. And a lot of it to me is just growth, right? I think the hyperscalers that do it themselves, I think it's easier to do when it's a smaller environment. But with the exponential growth, I think the hyperscalers that do it in-house -- well, I know that the hyperscalers do it in-house are struggling with it, and they're trying to find a better way to do it. So I hope that answers your question.

Do you see more competition in the future? Obviously, there's sort of a 3- to 5-year kind of barrier to entry on getting in this business?

I think so whenever there's money and you guys know this, whenever there's money in the water, the sharks are going to come, right? And so we certainly see people trying to enter into it, but there's something to scale with what we do. We're very, very embedded with our clients, right? It's not just the services we offer. It's -- we're embedded from an IT standpoint. We have APIs that go back and forth from a compliance standpoint. So we've done a really, really good job. On one customer in particular, we have individuals inside their data centers performing some data eradication. So I think we, as a company, have done a really, really good job embedding ourselves to sort of protect ourselves from competition. So I worry about competition every day. That's part of my job. But again, I think we've done a really, really good job making SLS sticky with the clients.

Has there been any pressure on the 70-30 split? Or has that been static over time? And do you foresee any pressure on that?

Well, no, at the moment, no, 70-30 has been fine. We do have other clients that aren't at 70-30. But overall, the average is 70-30. And Sean, you might want to speak to that.

Yes. As I mentioned, again, it's 70-30 to the client's benefit. So first of all, they do receive the lion's share of the revenue. In my experience and what we see now is if we can continue -- it's a rounding error to a lot of -- we never want to take anything for granted, right? So we always want to be competitive. But if we can continue to deliver the highest price, right, by finding the best market for this material and if we can continue to service our clients, we don't see as much pressure. I haven't seen it yet. Now it could happen tomorrow, but I doubt it because we offer such a great service that 70-30 really seems to be the standard. So...

And just on Ireland, so is that hyperscaler you've entered that market with in-sourced before and they've outsourced to you. Can you give us some color?

So they have a supplier currently, but they can't meet the capacity. So they've asked us to go move there to help with the additional capacity.

In addition to the program that you saw in our plant today is unique to SLS. So the hyperscaler, in particular, would like to replicate that process. And because of all the QC demand, the 3 years that Ingrid mentioned, it's not easy for the other supplier to get into the space. So it's much, much easier for us to walk in and do the redeployment and that way, they don't have to certify a whole another vendor.

Okay. So you've got guaranteed supply there?

No, nothing is guaranteed, but we have a very -- we have a line of sight to the volume, and we're very, very confident that we will be able to action on that.

Yes.

Just on the chart that you had up very early on around the dollars spent on data centers. Is it sensible to link that to your gigabytes sold, like there's obviously some mix and a few other things going on. But why -- I mean, I guess, why put that chart up? And if it doesn't link -- if it won't link to your volumes on a lag basis, why not?

So guidance question. So I think we put the chart up for a couple of reasons. Firstly, just to show that the amount of investment that is going into it. And yes, I think you can infer from that, that we believe that we will grow with it. The point that Sean made is that the stuff that we're now putting -- taking out the stuff that got put in '22, '23, not huge growth there. By the time we get to '27, '28, we'll be taking out the stuff that's in '24, '25, '26. So you can see that growth. So absolutely, we think it's going to be linked from a volume perspective. Our job as SLS is to make sure we get that increased share of the pie.

And it's held in prior years, like '22, '23 makes sense with your point.

Yes, it does. And there's no -- I mean, I don't think -- in global picture, we're not global in the total picture, there's no reason to think why it would change. what will change is what's our share of it. And so we think we're doing a very good job of it. If we just maintain our share, our percentage share, we would grow with it. If we can increase our percentage share, we'll grow faster than it.

And have you done any more work internally around where you think your share sits in the past, it's been very, very small for being the #1 player.

We have -- Ingrid and I were talking about that the other day. It's still -- I will hand over. It's still pretty small where we are because Sean noted that there was something like 8 billion...

8 billion gigabytes install of DDR4. So DDR4 for all that we've seen in literature, there's 8 billion gigabytes that are installed.

So in FY '26, we're forecasting to resell 1% of the 8 gigabytes. So it's -- we're still I guess implicit within your question is how much runway is there? I think there's still plenty of runway. We've got to win it. I do need to stress that. None of this is guaranteed to us. Okay. I might wrap up here. If -- there doesn't look like there's any further questions. Thank you, everybody, for joining us today. I thought an excellent tour. As you can tell, we're extremely excited about the SLS business. It's very complementary to our metal business, but I like it independent and complementary. If you haven't got the message that we think it's capital light, we think there's plenty of growth. I'm just going to reemphasize that message. It's ours to be got. It's a competitive market, but we are genuinely very excited about this business. Last point I'll make, we've been in this business for a long time. This is not overnight success. This is just overnight recognition. That's it. Thank you for the last couple of days, and we will see a number of you back in Australia.