Wednesday the 25th of April, 2018, and we would like to welcome you to the Asetek CMU 2018, and also at the same time, the earnings call for the first quarter of 2018. We're coming to you via the web from Aalborg in Denmark, and we're joined here in the room by a group of investors and board members and bankers and analysts and media and advisors and what have you. A good group. My name is Peter Madsen, I'm the CFO. I'm joined by André Sloth Eriksen, our CEO, you will meet him shortly, and our COO, John Hamill. You will meet him shortly also, of course. The objective today is to go a little bit more in depth with the structure, the fundamentals of Asetek. A little bit more in detail than what you normally hear at the quarterly earnings calls.
Let's take a look at the agenda. Legal disclaimer, very interesting, you should read it. There's an ice cream for the ones who find the spelling errors. The agenda today is the following, André Sloth Eriksen, he will take over in a few minutes, talking about liquid cooling, how it makes sense in a data-driven world. John Hamill, our COO, will come and talk to us about our markets and our business segments. We'll have a short break, where we will show some products here in the room, and there'll be some refreshments, too. We have some data center products, and we also have a more or less live version of a desktop application that you can see. It's not going to be so interesting for you following us via the web, but that is what it is.
12:30 P.M., I'll come back talking about financials and creating value through profitable growth. André will sum up at around 1:00 P.M., 1:15 P.M. At that point, we can open up the floor for questions. If you are following us from somewhere in the world via the web, then the application that you're looking at right now will allow you to post your questions there and we will see them, and we'll gather them from the iPad here. Here in the room, we'll find a microphone and have that passed around. That is the agenda. After today's presentation, for those of you who are interested, we'll make a factory tour, so you can see what it is we are actually doing here in the facilities in Aalborg. With that, André, you are up.
Yep. Hello, everyone, and welcome to Aalborg. Can we crank down the speaker sound a little bit? Before I'm going into the details, I would just like to make it very clear that for those of you who want to stay until 4:00 P.M., you are more than welcome. It seems as three people already told me that people are speculating that since we changed the schedule from 4:00 P.M. to now one something, that's a negative, and the stock is going down. I'll just make sure that all of you can stay till 4 as originally invited for if you want to. Next time, I have this smart thing here called a phone. If somebody is in doubt that whether we cut the schedule is because of something negative or something positive, just call me, then we'll figure it out.
Back to the agenda here, liquid cooling makes sense in a data-driven world. Today, you will hear some good stuff, some not so good stuff, and some surprises. That's why we are here. For those of you who are new, probably not so many in the room, but perhaps on the web, what we do is pretty much one thing. It's liquid cooling for data centers and for PCs. I'll just stand over here for the sake of the camera and for the sake of you. The way we are organized is that we have offices in the U.S., we have offices on the West Coast, where I was myself for almost eight years, to be closer to the customers. We are in Texas, also to be closer to the customers who reside there, namely Dell and HP.
We are in our headquarters right now where we have our R&D facilities. We have some manufacturing as well. Our main manufacturing is in mainland China, in Xiamen. We have sales offices in Taipei as well. We have essentially two business segments. One of them is desktop PCs, and the other one is data centers. The markets we try to dive into right now, especially on the desktop side, it's gaming, it's esports, it's virtual reality, et cetera. On the data center side, the markets are HPC, so high performance computing, just like the order we announced this morning, that's for an artificial intelligence supercomputer in Taiwan. It's servers, server racks, and of course, also hyperscale. In the demo later today, we will show you some of the stuff if you're interested, of course.
If we should talk about recent highlights, more or less real time, our Q1 revenue went up 21%. Our desktop business is still booming, I would say, it's now the third or fourth consecutive year. We surpassed a total of 5 million units shipped now, so 5 million units deployed in the field. We announced with Intel PCSD, and as we'll get much more back to, and as I'm sure the stock market has realized, that has not really met our expectations. That was the bad news I talked about just 30 seconds ago. That's of course annoying and frustrating to all of us, but I'm sure when you see it in a bigger perspective, and you see what we see, then I think there is still, what should we call it? There's still hope for that deal.
I would also make sure you understand, and John will get more back to that the Intel deal for now is two different things. There was original, what should we call it? A run rate business, and then there's the RackCDU business as we have over here. That's still very much alive, but because it's just new and just announced, it will take some time, just as it has done with Fujitsu, for example. An interesting slide here, that we've been working on gathering the last few days is, after we went public, the desktop market has boomed, and the desktop business has boomed for us. How did it actually look when we started the desktop business?
It was actually very much the same as what we're seeing on the data center market. It actually took us pretty much 8 years from we started the desktop business until it really took off. As you know, this is not science. This is not something I can say it's going to be exactly the same on the data center side. I had hoped it would go much quicker on the data center side, but at the end of the day, there's a lot of similarities in the sense that liquid cooling was new into both markets. It was pretty much the same customers we went after, not the same market segments, but the same customers, so the big OEMs. I think what this slide shows more than anything is that it's really volatile, and it just takes a lot of time.
I think that's an important slide. Perhaps the most important slide that I'm going to show today. Why is it that we are selling liquid cooling, after all? Why is it people are buying it? That's because we actually have some strong value propositions in both segments. That's also why I'm, let's say, relatively comfortable on the data center side because I think we have so strong value propositions that it is a question about time. Some wise man said that in the long run, we're all dead, and that's of course true. I do believe that the data center market will turn around. If we look at the data center side, as I just talked about, we can lower the cooling cost in a data center by up to 50%.
We can increase density quite a bit. We can increase performance, and these are all very important factors in the data-driven world we are in. John will get back to that in a second. On the desktop side, I think it's pretty, I would not call it crazy, but I think it's quite substantial that we have actually sold more than 5 million units now. That's quite a bit. There's more than 5 million gamers out there with our stuff inside. People in the gaming world, they buy it for a better experience, they buy it for lower noise, and they buy it because they can overclock. The setup over here is actually a virtual reality setup, and I'll invite those of you who are fresh to actually try it out. It's quite fun. We have a small video now, I think.
We're consumed with the need for data. We want it fast. However, our need to know has data centers working nonstop, consuming annually over 1.5% of electricity used worldwide. It's literally a hot business. In the U.S., over 86 billion kilowatt hours are needed to cool the servers holding the data. Fortunately, there are people trying to fix that. With Asetek's RackCDU technology, power needed to cool data centers could be reduced by 50%, and it could decrease overall power consumption by 25%. With that saved energy, we could power 4 million homes annually. Energy savings from RackCDU means more resources for everybody. If every U.S. data center used Asetek's technology, we could save enough water to fill 63,000 swimming pools and prevent the burning of over 8 billion pounds of coal.
RackCDU can reduce worldwide electricity costs by over $7 billion a year. That's more than the GDP of several countries. Best of all, these savings can be recouped by data centers in less than a year. That's liquid cooling done right.
Some of you may have seen the video before. It's not something we just made up for today, I think it shows very clearly the value propositions I'm talking about and why I at least believe that the data center market will come around. Granted, it was good timing that we got the order that we released this morning. I think we will talk more about it, John, as well, the whole HPC side and artificial intelligence side of things. I think these are just the early adopters, just like we had the early adopters on the desktop side. Just to furthermore emphasize why this is taking time, it's not uncommon that we spend one to two years with any new OEM or new partner, just back and forth, sending prototypes, making sure they understand.
They have their own testing and development, which can easily take one to three years. We have revenue and product launch, and what we've really been caught up in where we are right now, is that all of this is true. This is also true. However, in this case, the customer's performance in the market just did not meet our expectations, and that's the risk that we are running with these customers. If you look at our customer base right now, if I should highlight something here that's important and has been important for the company, I think if we look all the way to my right, HP in 2007. If we go back to my other slide of how long time it would take, that's when we announced HP for the first time and the volumes and the revenue ended up being massively disappointing.
However, if we now look back to 2007 and look at HP, they have been a great customer for us and is a great customer. I think that's the way we need to look at it. If you look at our data center up here, you can see the data center OEMs that we have added. You can see here on the desktop side that we are continuing to add new customers, and that is how it will play out up here. I also have to stress, I feel that the data center side is a venture still. It's still new. Our desktop business is solid, it's growing, it's performing over expectations, but we are not there yet on the data center.
I at least, as the CEO of the company, think I need to put a line in the sand and say, "This is where we are." Although we would like to, we cannot fix it one quarter over the other. That's the picture as it is right now. I'll not spend much time on this because I already addressed it. I think much more interesting is if you look at our revenue, if you just forget the composition of the revenue for a second, but actually look at the revenue. As you can see, our Q1 revenue, it's by far the largest Q1 that we ever had. As some of you remember me saying, we always have one bad quarter in the year. We just don't know which one it is.
Traditionally, it's one or two, and without getting ahead of myself here, I think John will talk a little bit about it. We have made this gray bar here to the right, because from what we can see now, and there is a disclaimer in that because the world can change after this meeting, but from what we can see now, our Q2 is going to be a killer quarter. It's going to be really strong. It's not all bad, let's put it like that. What we've also done for today is that we have actually looked at the markets again. We looked at the markets we are in at the IPO and when I was at the IPO roadshow.
Since then, we've had a little bit the attitude that, okay, these markets are so big that we don't need to spend a lot of time noodling around figuring out if it's $800 million or it's $700 million. We've done that, and we've confirmed that. I think what you'll find is one of the biggest surprises perhaps actually how big the desktop opportunity really is. When we went IPO, I positioned our desktop business as something that was a short-term cash cow. That's something we need to survive to get into the data center where the real money is. I think that story changed a little bit, and you can argue that's convenient now that we have an underperforming data center business. I acknowledge that it's a convenient argument.
On the flip side, we have now been knocking it out of the park for four years straight on the desktop business. I think the numbers speak for itself that the desktop part of our business is actually a real business, and it's a very healthy business. That does not excuse that we are underperforming on the data center side because I want to fix that. I just think from an overall perspective, that we have a solid business that's really growing, which is nice. That leads me to say that we are in large and growing markets. We are supplying some of the best and most respected brands on the planet. Case in point, this morning's press release from Taiwan, we now have, well, have, that's perhaps a little premature, but supplying the two largest supercomputers in Asia, so in Taiwan and Japan.
I at least would like to believe that they are not crazy people in Taiwan and Japan. They are just ahead of the curve compared to the rest of the world, and I think the rest of the world will follow. I will not talk a lot about IP today, but obviously emphasize that we have our strong IP position. Because of the desktop business, we actually do have profitable growth. We are building the data center business with the money we earn on the desktop side. To me, that's important because pre-IPO, when we were only a desktop company, we were actually living by the mercy of our investors. Thank you to those who are here, still around. Where today it actually is controlled and paid for. I think that's nice.
Before I leave the deck to John, I would of course like to know if there's any questions. I will be back. I will sum up, et cetera, so you will see me again. If there's any clarifying questions to what I've said, then please come ahead.
If that's the case, André, we just need to source a microphone.
We can do that coming up.
Yeah.
Let's do. Are there any questions? We can perhaps figure that out first. I don't think there are any questions, Peter, we can move on with you, John. All right.
I'm going to try and put some flesh on the bones here and add a little color to some of André's commentary, and hopefully I'll manage not to contradict you. I thought I would start with some macro-level trends, really just to set the scene. Let's start with data center, because what we see in the data center market today is lots of new, very interesting applications. You might be familiar with some of them. You might have heard of big data analytics. I'd be surprised if you hadn't heard about artificial intelligence. One part of artificial intelligence is machine learning. You might understand that as autonomous driving. There's a copious number of new applications coming to the market, and they have this insatiable desire for data. Huge amounts of data.
Data requires processing, and a lot of this data requires what we call real-time processing or near real-time processing. That means very powerful processors, and that's very good news for Asetek. Okay? When you're thinking about the macro level, all these new applications, these new technology areas that are coming to the market, very good news for Asetek, and it's not confined to the data center. In the client space, or the desktop space, if you prefer, we see end users chasing this immersive experience, this desire for their game to be as realistic as possible. That is pushing the boundaries of gaming hardware on a daily basis. You'll get a little chance to try it for yourselves, and if you have never tried virtual reality, I really do encourage you because it'll give you a flavor for what I'm talking about here.
With that, let me continue here. I'm going to jump into desktop first. Really our desktop business today is dominated by two consumer segments. We call one DIY, we call the other OEM, but it's an abbreviation. You can see the full names down here. Our DIY business involves us selling liquid coolers to brands like Corsair and NZXT. They take our brand, and they sell it directly to end users who install them in their own computers, their own gaming computers, their own gaming rigs. The other part of our business is OEM, where we sell to companies like Dell, Alienware, and as André touched on, HP. Those guys sell complete systems to gamers traditionally. Okay? Now our business is split very roughly three quarters DIY, one quarter OEM.
Last year, we managed to ship 1 million units into the desktop segment, and that trend continued in Q1. What does that mean about our penetration in the market? How much market is out there, and how are we doing? We think our penetration is below 10%. Our logic is based on the rationale outlined here. We know the high-end gaming market is roughly 22 million individuals. That means 22 million people have paid more than $1,800 for a gaming PC. More than $1,800. We cannot count notebooks. Notebooks do not employ liquid cooling. I got to take that out, guys. I have still got a population of north 12.5 million, probably nearer 15 million, that I know are refreshing every three and a half years. That is about 5 million units a year that I can sell into. That is only the high end.
There is at least as many in the mid-range of the market, as many customers, as many end users. If I aggregate what I see in the high end and what I believe is in the mid-range, we see at least 10 million units, which is why we believe we have less than 10% penetration. This also gives us an opportunity to size the market, because I know from last year that my ASP was just over $50. That tells me that the market opportunity here is just over one half billion dollars. Okay, that was what André was referring to earlier on. This market really did surprise us when we went back and sized it. The other interesting statistic, excuse me, about this market is the growth. We actually see very healthy growth here, around about 7.5% compounded annually. What is driving the growth?
For those of you who heard my comments last year, there is no new news here. Okay? We know aspirational display technologies are a huge part of what is driving the growth we see in the gaming market. As André referred, virtual reality is a big part of this. Let me make one thing clear. The attach rate, meaning the number of computers that use virtual reality is very low. It is single digit percentage. What is driving the growth is that a lot of enthusiasts and gamers think at some point down the line they are going to use virtual reality, so they future-proof their purchase. They buy a bigger graphics card, a bigger processor than they think they need. Okay, virtual reality has this very interesting effect on our market. Probably the biggest driver are games themselves.
We see very roughly, guys, 10 to 12 games a year which just send the gaming market nuts. We call them AAA games. Of course there is esports. Esports is this whole phenomenon that has emerged where people go crazy watching other people play games. Last year we talked about esports as providing a steady stream of new gamers coming into the market. We really could not quantify that benefit, but we did some more research and we think the impact of esports is measured in hundreds of millions of dollars. Indeed, by 2020, we believe that the impact of esports on gaming hardware purchases will be three quarters of a billion dollars. Quite a staggering amount of money. Let me dive into one more topic before we summarize where we are at in desktop.
I want to look at just how our business is spread across our top five customers. I list our top five customers there. You can see them, Alienware, Corsair, EVGA, NZXT, and of course Thermaltake. Consistent with our split between DIY and OEM, there is only one OEM customer up there, and that's Alienware. The other four are all DIY customers. They contribute in the aggregate about 80% of our desktop business. What I've tried to show on the graphic here on the right-hand side of this slide is just how that 80% is partitioned between 2016 and 2017. I think it's fairly obvious to see that there's quite an encouraging trend there. We've managed to see our business spread more evenly over those top five customers, effectively reducing our dependence on our biggest customer. Strategy and outlook. We want to continue to dominate.
Of that you can be in no doubt. To do that we're going to drive innovation. We have to drive innovation. We have to have the best products. We also have to drive differentiation. If my customers cannot make money, I don't think that's a solid business case. We're very focused on enabling differentiation for our customers. One that I touched on last year, we do have to increase GPU attach. There's a lot of challenge in there around the education of the market that we've yet to overcome, but it continues to be one of our key elements of dominating in this space. In terms of the outlook, as André said, we are looking for at least 50% growth year-on-year in Q2. However, it's too premature for me to give any more. How do I put this right?
I have very limited visibility into Q3 and no visibility into Q4, my guidance for the year will go from 5%-15% and increase from 10%-20%, again, for the reasons being I have very limited visibility just now into Q3 and no visibility into Q4. Okay? Let's move on to data center. Data center, I think the key message here is that our value proposition resonates best in the HPC space. HPC is where we're seeing all these new applications I referred to earlier on, big data analytics, artificial intelligence, in whatever variety it shows up. We're seeing these new HPC applications across a myriad of verticals. Used to be HPC was really confined to government labs and education. Not anymore. Showing up everywhere.
We actually think that the segment is now around 18% of the entire data center business, which is very encouraging, because again, that's where our value proposition resonates most. It's also growing. That should come as no surprise, right? I mean, all the hype we hear around all these new technologies, I'd be surprised if it wasn't growing. I'm actually surprised it's only grown by 5%, and that's 5% compounded annually. What about the opportunity? We think it's vast. Vast. We estimate around about $800 million in this past year, and that jives fairly well with where we were back in 2013, right? I mean, we knew it was hundreds of millions. We think around $800 million. Based on last year's revenue, that says our penetration of the market is less than 1%. Huge opportunity here, guys. Huge opportunity. Our strategy, well-publicized.
Win more business with our existing OEMs, add more OEMs. Since we last spoke, we've added six, most notably Quanta this morning, which I'm delighted about. We added Intel, and that was another hard-fought victory. Let me just add a little bit of color here on the Intel collaboration. This collaboration began more than 2 years ago. Initially, we focused on what we call sealed loop. This is a sealed loop in here. The entire liquid cooler is in this assembly here. This is where we focused our energies originally, and actually, this is the product. As I say, we started that about now more than 2 years ago. More recently, our engagement with Intel has increased and broadened to include our RackCDU direct-to-chip technology. What's the status on both those efforts?
Intel has now launched this product, and if I refer to it as a LAC, I'm using Intel internal language. LAC is a liquid-assisted air cooler in Intel parlance. They launched in March, which means Intel's sales and marketing, Intel's technical support are now talking to Intel's customers. Intel, in this instance, is Intel PCSD. An important thing to note here is Intel PCSD sell motherboards. They sell node trays. They sell chassis. They sell racks. When I talk about Intel sales and marketing and Intel technical support, they're out talking to data center OEMs, system builders, system integrators. Okay? The challenge we have, as André touched on, is that the forecasts and the orders that we're now seeing are not meeting our previous expectations, and those expectations were substantial. Let me try and quantify that for you.
We'd actually hoped we would see annual revenues measured in millions. It's beginning to look like those annual revenues are more likely going to be measured in hundreds of thousands of dollars. Okay? On a more positive note, as I mentioned, we have engaged on our RackCDU direct-to-chip technology, which means very shortly, Intel will be talking to their OEMs, their system integrators, and their system builders about this technology. That's great news for Asetek because it means our reach into the market increases enormously. Trust me, Intel has a lot more resources than Asetek. That's consistent with our strategy. Think about what we've described in the past. We've talked about winning more business through our existing OEMs and adding more OEMs. They're really going to help us with that second part of the strategy.
Actually, we're already starting to see some benefits from this engagement because our pipeline is starting to show some Intel OEMs and system builders, or at least some opportunities are showing up from Intel OEM system builders, system integrators, and some of those are new to us. Having said that, it is very early days here, and it's probably premature for me to try and predict what the impact of this will be. Let's talk a little bit about resetting the guidance here. As I mentioned, now that we're seeing orders and forecasts, it's become apparent that the demand we had anticipated is not materializing. I've tried to illustrate the impact on this graphic. What we have here is our 2017 actuals. Our 2018 initial guidance with this gray block representing what we had hoped for sealed loop and where we are today.
A few comments on the current outlook. The OEM section of the forecast is dominated by pipeline opportunities. A pipeline opportunity is a specific cluster that one of our OEM partners is going after. The Quanta deal at NCHC in Taiwan is on that pipeline, and that'll give you an idea. Think of very specific orders. We, over the years, have developed enough experience and expertise that we can distill a pipeline into an outlook, but it's not an absolute science. Okay? It is a lot of guesstimates and estimations. The other thing to note is that this current outlook only reflects opportunities that we can see today. Clearly, if there are more opportunities in the pipeline, and thereby the outlook may change.
Having said all that, we feel based on what we know today, the most likely outcome for the year is going to be flat growth year-on-year. It's an oxymoron, flat growth. Let me just conclude with this slide. Our goal is pretty simple. We want to create a meaningful business, meaning meaningful revenues and a profitable business. Don't think I need to say it again. More business with our OEMs and additional OEMs. I think you've heard that. We are in a very good position. Our value proposition is very strong in HPC. We are in a very good position, and we have to exploit that. The one thing I want to stress before I sit down is, despite the reset on expectations, we are very committed to this market, guys.
If there's one takeaway you have from my comments in the data center section is, it is a huge opportunity. We're very well-positioned, and we are very committed. With that, I'll pass the bat on.
Yeah. Maybe we should open up the floor for questions at this point for your segment. Yeah. Let's get a mic. Yeah.
No problem. Peter, you need to aggregate.
I shall do my best. I can see there's a gentleman.
Facility
In the middle of the room. Mr. Jens. Sorry about that. There you go.
Jens Ole from Danske. If you look at your guidance, do you see, is that then without any CPU full rack visibility from Intel? You said you saw something happening in the pipeline. Does that actually include anything there?
Yes. If it's in our pipeline, then we have a way of aggregating what's in the pipeline. The outlook does include some of those Intel opportunities, at least the ones we can see.
Okay. Thank you.
If you could pass it on to Anders from SEB.
Yeah. It's Anders from SEB. Talking about the GPU attachments, can you add some more flavor to that, and is that included in your market calculations for the desktop business?
Yep. Our sales are disappointing. For GPU, we had a lot of ambition. We discovered the phenomenon, and you heard me reference the education. We have yet to figure out how to educate the market. The phenomenon we've discovered is this. If you have a good, better, best GPU card, I can sell you a good liquid-cooled card, or you can buy the next one up air-cooled. Customers would rather buy the next one up. Come the next one up. What we see is customers would rather buy an air-cooled card than a liquid-cooled version that's not one generation, but slightly lesser performance. We hadn't anticipated that. The challenge we have is that we have to educate the market on why it would make more sense to buy the liquid-cooled version.
That's much easier to do in a VR environment than it is in a traditional gaming environment for a whole bunch of technical reasons that I touched on last year. We're struggling to get that message across, in part because VR attach rates are as low as they are.
I just have some flavor to add also.
Yeah
that I think is important. When you buy a CPU, there may be a cooler that's in the box, but it's not actually attached. When you buy a graphics card, the cooler is attached to the graphics card itself, and it's actually stated that if you remove the cooler, the warranty will be void. That is a topic as well. Where on the CPU side, it's more for the do-it-yourself users to go and put on another cooler. On the graphics card, that's still, let's say, a challenge. The most hardcore guys obviously do it, and as you can see the one we have here, if you'll see later, it's also liquid-cooled. There's nothing to stop it from a technology perspective, and I would say the value proposition on graphics cards is actually higher right now than it is on a CPU.
There are some boundaries there that's also a hindrance, but I think we can work through those. That was the one thing I had to say. The other thing, John, is in terms of the pipeline. Could you just elaborate a little bit about how many opportunities do we have in the pipeline? How do we categorize them?
Yeah.
Why is it that you got caught by surprise, a nice surprise that is, about this order that we received this morning?
The way the pipeline works is we're constantly communicating with our OEMs, constantly talking to them. Think about it this way, if they are working on an opportunity, that's a potential supply chain challenge because as soon as they ink a deal, there are penalty clauses if they're late or delayed. They're motivated to talk to us and tell us what they're doing. We have these discussions going on with all our OEMs, constantly chatting to them. We're constantly saying, "What are you bidding on? What does it look like? How big is it?" I need to know that because I need to be ready to supply that. Right just now, based on conversations we have ongoing with all the usual suspects, there's about 30 opportunities in our pipeline.
Based on the conversations we're having, we rate them as high probability, medium probability, and low probability. We use that probability to factor the revenue, if you like. We do some funky math on what we think it's worth versus the probability of landing it. That's an ongoing process. That's something the Asetek sales team and their customers are constantly doing. We have a tool that we use to manage all this. It's a very difficult process because I'll give you a for instance. We have one opportunity that we thought we'd be talking about today. It got delayed. It got delayed because the data center's delayed. There's nothing I can do about that, guys. There's nothing my OEM can do about that, and that's not uncommon.
Maybe some other part of the supply chain's messed up, there's no point in buying all the processors and all the racks and all the liquid cooling. It's very, very difficult to predict. Very difficult indeed.
Could you then elaborate why it was a surprise that we received the order today since it was already in the pipeline?
I'm not sure you'd ask me.
Well, I can do the answer then because I know.
Yeah.
It's not like we get a month warning from an OEM saying, "We will be winning this deal," or, "We won this deal," then we can prepare for it. The facts here on this order we announced this morning was yesterday afternoon in our board meeting, in the very board meeting, John gets an email saying, "Oh, positive news. It seems as Quanta has won this deal." Obviously, we cannot report that we think that we won something, we have to wait. This night, in the A.M.s, we got the confirmation. We had 12 hours of warning. That's what we get.
Yeah.
That's why it's difficult, and that's of course also why the trick is that our pipeline, let's say we have 30 opportunities, we need to have 300 opportunities before we can turn this into a steady business. That's really what we lack.
Yeah. I see. I understand the question. What was quite funny was for that particular opportunity, we had more than one customer bidding, and call them sore losers, but some of the other OEMs were saying, "Nah, they'll never win that." We had a counter-opinion, which just to confuse things. Yeah, that's the other thing. We do often have more than one customer going for a bid, so we can get different views into the bid, and we see what's happening from different perspectives, which does help.
Just one more thing that I think people are interested in knowing, John. Out of the 30, are there any of the opportunities that we can actually credit the Intel collaboration for?
About one third.
That means 10 in English.
About one third. In that range.
It's not that this Intel partnership means nothing. It's just that that part of the business is bid-dealt, or bid-based. I'm sure when we look back at this a few years out, it will be the big success that we all hope for. That was kind of the questions that I wanted to facilitate, John.
André, just following up on that. You say you have 10 projects roughly in the pipeline because of Intel. Have they basically come into your pipeline after the mid-March announcement?
I think both before and after and during it. It is very much a real-time process. There are, of course, projects in there that if they materialize, and we hope they do, then you will all see, "Now we understand." We just cannot really get ahead of ourselves yet. As John said, there are opportunities that we had expected to be able to talk about today, that we are not. On the other hand, the one that we are talking about, we had also not expected. That just shows you why it's difficult.
Essentially, your pipeline has gone up materially because of that announcement.
Not because of that announcement, because of that technology. The announcement in March was this technology. That was what's been launched.
I think a misunderstanding that we can also clear up here, Intel is a gigantic company, and if you want to know anything about Intel, you have to ask Intel. What I can say in relation to us is that this technology is not something Intel is going to put up on their front page and say, "Look what we just did." This technology is something that they will sell to their customers, and their customers are not you or your cousin or anyone else. These are system integrators and OEMs. That communication is very much direct. The fact that we've been working on this thing, the OEMs and the system integrators, they've known about that for what, 18 months or something like that.
Do you see any change in size of orders in the pipeline now? Looked a few years ago, they were very small. Are the size increasing and anything changed with the Intel pipeline?
I'd love to give you a clever answer, but. We've tried this.
The business is too small just now. If we had more wins to aggregate over, we'd probably have a better chance. We've actually had one or two massive wins that skew everything one way. I think we need to see more wins and it would give us a better chance of aggregating. We could perhaps aggregate across the entire business and guess what an average win was like. We've struggled with that, I'll be honest with you.
Too early.
It's part of the unpredictability problem we have, which is, hey, what if we won six deals, what would that be worth? Or no.
On the flip side, we have won a couple of smaller deals that we did not announce, that we probably would have announced a couple of years ago. For sure it's moving in the right direction. With only 30 opportunities in the pipeline, if we get them all, it's massive of course, but we can't really apply any statistics on it yet. We need to wait a little bit longer to be able to do that. For sure, there are deals all sizes. The deal we got today, I consider that a decent, a good deal.
I think another data point that's perhaps important is that, if we should talk a little bit about one of our other customers, Fujitsu, that surprised me at least, that what they've been selling in HPC the last year or so, apparently seems to be with liquid cooling, more or less all of it. These orders that may not occur as the world's biggest orders to us, they are quite material and significant, the orders for our customers.
Yeah.
Thank you.
Very good. Maybe Jens, if you want, you can hold onto the microphone, maybe turn it off. We have a few questions, John, that came in via the web, if you are ready.
Be delighted.
One of them is from an analyst in Oslo. We are now back in desktop space. He's asking, "You noted at least 5 million units for the mid-range as the mid-range gamers addressable market for liquid cooling. Have you conducted research suggesting that you are seeing material uptake among mid-range computers? Also, please specify further what you consider mid-range other than the price and processor. For example, processor types.
Well, price is exactly how I would define it. I'm not sure I know how else to define it. I would say mid-range would be very roughly between $1,000-$1,800. I think at that price range, you can easily accommodate a liquid cooler. I think it's much bigger. A much bigger segment, and I think the assumption of 5 million is extremely conservative.
I think we can say, John, that for those not looking at the slides, but this study about the desktop part of our business, it's not something we just pulled out of our behinds. It's something that we bought from Jon Peddie Research.
Yeah.
That I actually also used when I went on the IPO road show and when we made the material for that, so there's some level of continuity in it. He is probably the best one to talk about high-end graphics, high-end gaming PCs, et cetera.
Yeah.
These are based on data from him.
Perfect. Continuing, GPU uptake over the future 6 to 12 months past that point in time, what do you see there?
I don't see any significant change just now, just because I'm not sure that we're going to see any improvement in our ability to educate the market. That's a challenge we have to address. I left it on there deliberately. I highlighted it as an issue, but that's still a challenge for us.
Very well. We go into a question from Germany. How are you holding up your desktop lead? The desktop lead. How's that going? Holding up the leadership position, is what I read here. How long are your patents valid, et cetera?
Patents, 2029.
Yeah.
I don't think it's about patents. I think it's about providing the best products and there's different ways of looking at the best products. You can look at performance. Performance in the desktop side means thermal performance, which I won't bore you with, and acoustic performance, meaning how noisy it is. There's also quality and reliability, and I think there's no doubt that Asetek is the market leader with some distance in that space. I think there's different vectors you can look at. The challenge we have is we want to stay there, and that's why I highlighted the innovation.
I think we need to continue, we need to double down on making sure we're the performance leader, making sure our quality and reliability is what our customers expect.
Sure. Talk about Q2 desktop. You opened the kimono a little bit.
Yeah.
How is HP and volume looking? Can you talk about that at all? Why is it that we see a strong Q2 to HP?
I'm not at liberty to discuss. There's some stuff going on that I'm not at liberty to discuss at this point in time.
What we can say, just to get that out of the question, I've learned now the hard way, I think that investors speculate a lot.
No.
What I can say, it's not by giving extraordinary discounts, let's just rule that out, that we have not fabricated a strong Q2.
Sure.
This is the real demand that we see.
Yeah.
Let's see here. In your opinion, did OEMs not want the closed-loop solution to the degree you expected?
I think we should not frame it like that because this has nothing to do with liquid cooling. Let's put it like this. We rely on the success of our customers. If our customers do not sell their product. As you know, I love to look at the car industry. Let's say that we made massage devices for car seats, like I used to have in one of my BMWs. I was probably the only one in Denmark buying it. If we had been in that business, then it would be very much like a pick option. If you were actually the manufacturer of all the car seats in a BMW 7 Series, we would sell exactly the number of seats that they would sell cars, and that's where we are right now.
This does not come down to whether the OEMs like closed loop or not. This is a question about what products do the OEM wants for the CPU or GPU in this case. Of that, we are completely without influence.
Very well. There's a question from Switzerland, I believe. Let me hear it out. What do you see as the main reason for the disappointing uptake in sealed solutions sold by Intel? What needs to be changed by Intel to improve performance?
That's a question for Intel.
Perfect.
Yeah, I think I just answered it.
Yeah. Going back to Germany. Isn't this a good chance of a strong acceleration of the data center business in 2019 and the following years, given the much higher potential of rack cooling? More for André.
In theory, yes, because we opened our channel quite a bit with Intel, of course. I'm also a realist after having been a world champion in guessing wrong in the data center market for the last three years. In theory, yes, but we'll take it one quarter at a time and see where we end up.
Perfect.
I think Johan had.
If I may, just one final one to finish.
Sure
Going back to desktop and graphics card, how big is the opportunity of graphic cards of cooling in the future years, GPUs? What's the term there?
As André said, the value proposition is very, very strong. As André also said, there's really not an aftermarket. People generally don't rip apart a thousand-dollar graphics card and slap on a liquid cooler. To the extent that we can convince the manufacturers of graphics cards to do liquid cooling, there's a reasonable market opportunity there. If I was to try and quantify what might be a good target, it would be 10% of the volume we do on CPU coolers. That would be a good target. As I say, we've got challenges getting there.
Sure. Johan Danske Bank.
Thank you. Just a little bit more on the desktop Q2 strong expected growth. Why are we seeing it this Q2? Normally Q2 is not the biggest quarter. Is it new product launches or what drives the growth?
To you or should I?
Yeah, there's going to be some new product launches. I'm not at liberty to disclose any more detail. That's part of it. That's part. It's not the only reason, but yeah.
We have seen a large order flow last year of smaller orders for the data center business. Why has they stopped? What are the feedback you're getting from these smaller OEMs that you have onboarded during the last year?
I think the reason why the news flow has stopped is because of that, not because of the order flow. As I just said, we've had a couple of smaller orders. I don't want to lash out too much on the stock market, but we got the impression that sometimes when we launched smaller orders, we got negative feedback because then there was speculation that, "These orders are just because they don't have any bigger ones to tell us about." We decided to not announce the smaller orders. We get the feedback that why don't you have them? Just the brutal, honest truth, this is more a question about investor communications than anything else. We would be happy to let you know every time we win an order independent of size.
Unfortunately, I can also say that, as you know, I go on road shows a lot, let's just say the feedback is absolutely not the same of what people want and what they do not want. It's very spread out. There's nothing I would like to do more than tell you about every time we get an order, because it always have some relevance. As some of you have heard and probably also understood by now, we cannot pick and choose. We report when we get an order. It's not like, "We don't have a big one. Let's send out a small one." It doesn't work like that. We get the orders, we communicate them when we get them. I think that's the best answer I can give you. I can ask a question back to you.
Would you prefer, as an analyst, that we did announce every order?
I don't know what I would prefer, but I would prefer to know that something has changed the communication strategy about it.
Which I know now.
Well, that is actually something Peter was going to bring up anyway, but we've just not made it to his presentation yet. That is actually something he would have bring up or will bring up.
The last part of my question, what is the feedback you're getting from all the orders that you have implemented during the last year, the small OEMs and such?
I'll give you a great example. NCHC just bought their second Asetek liquid-cooled cluster. They already have the number 95 liquid-cooled cluster on the TOP500, and it was that experience that encouraged them to move forward. I think that's a huge endorsement.
Keep in mind, our business model is not to sell to data center. Our business model is to sell to the OEMs. Of course, we are in dialogue with the OEMs, but we don't really speak too much to the end users. They don't come to us. They come to, for example, if it's Fujitsu, they will go to Fujitsu and work out whatever issues they have with them. I can guarantee you, if there's anything negative, we will hear about it. If there's something positive, the product works as expected. It's not like we have a lot of dialogue with the end user. We have to be cognizant about the fact that, as John said, we have different customers sometimes bidding for the same end user.
We don't want to step in between that process. We have to back off a little bit.
If all the OEMs are happy and we're seeing an increased penetration of liquid cooling in data centers in general, shouldn't you be expecting that the OEMs that you currently have would sell more during 2018 than in 2017?
Yes.
Yep. That's what we hope for. Doesn't work like that. Doesn't always work like that.
The OEM part of the guidance is a conservative one because you don't have visibility on it yet. Not that I'm putting it into the universe.
You are the analyst.
Of course.
You analyze what we say. What we say is what we see.
Okay.
I don't apply any value to it if it's aggressive or conservative or neutral. What I tell you is what we see.
Very good. A couple more from the web. I have been corrected by a Norwegian analyst. I rephrased this question the wrong way, so let me read it out here. Why, in your opinion, did the closed-loop solution, and I'm assuming we're talking Intel here, why did the closed-loop solution not sell as good as you expected? What was the main headwind, and how could this change?
Okay, to answer it from Asetek's perspective, the main problem is our customer doesn't want our product. The way it can change is if our customer increase their forecast.
Very good.
I cannot speak on behalf of Intel. I don't know how I can put it any clearer.
Sure.
If you have a question to Intel, please call them.
1-800-INTEL. One more timeline on RackCDU with Intel. Do we have any feedback on that or input on that?
We're already factoring it into our outlook.
Yeah. I have two more questions, then we'll have a break, come back here, and then we have a more formal Q&A session at the end of the session also. I have two more here. One from Germany, I believe. Does the agreement with Intel also include new products under development, and are you developing products exclusively for Intel?
Yep. We have a partnership agreement with Intel. We'll disclose details as and when it's appropriate. I think that's probably the correct answer.
Yeah. What I can say is, I can repeat that even if I could, I would not stand here and talk about Intel's future roadmap. That would be close to suicide. What I can say is that we typically do not develop products exclusively for anyone. What we do is that we work with each of our customer to make something exclusive for them. It could be features, or it could be something on their side. Typically, we don't do exclusive products.
Very good. One final one in this segment that does not pertain to Intel, actually. How do you evaluate your success of the in-house solutions of the large data center OEMs like Dell and HP? How do you evaluate the success of the in-house solutions of large data centers?
Okay, get it. I think what's been asked here is that Dell and HP have some in-house liquid cooling solutions, and without being arrogant, although it may come out like that, I have not seen them yet in any of the markets we operate.
Perfect.
Is that-
Yeah. I think HP essentially acknowledged the Apollo 8000 was a product with lots of challenges when they essentially abandoned that and moved to purchase SGI. I think it was
Well, I think the essence in the question is, do we see fewer orders because of what, in this case, Dell and HP has done internally?
I don't believe so.
No. We don't believe that at all.
Good. Perfect. Yes?
I just have a question on the desktop side. Now you have quantified the market you are in the high end. Would you consider actually moving down the value chain to get a bigger share of the wallet or?
No. I can give you a very exact reason why that is. One thing I like about our company is that we make money on what we sell, and I think I can easily disclose that with Dell, a few years back, we actually looked at a deal where we would move down the stack substantially and really be talking millions of volumes. The point was that to do that, we would have to move the price points down to a level where it was just not desirable. We would effectively turn ourselves into what we call metal benders. That's cost-plus business model. I think it would ruin the strategic value of the company.
I think it would ruin the strategic direction of the company. I think it will turn us into much more like a, let's call it Chinese ODM type of company rather than a technology company. Yes, the opportunity is certainly there, but it's not something we are considering doing. Finally, it also has to make sense for the end user. If there's not really a value prop other than saying that it has liquid cooling, I think we failed. There is a gray area in between, and a difference that I see in the market. When I founded the company, PCs were a big thing. Enthusiast PC, you could build your own PC, and back then we actually also had LED lights. I'm not that old, but still it's 15, 20 years ago, and it has all come back now.
If I go to a computer store here in Aalborg locally, I can see a lot of gaming PCs that are actually not liquid cooled because it's price. There is a gray area that we have, I would say, perhaps not exploited in full for several reasons. One of them, I can take the opportunity to re-emphasize that, one of them perhaps is that we have not taken the desktop market as seriously as we could have or should have the last couple of years because it's just been going. You didn't say that, John, that's something we actually agreed to do, is to beef up the R&D side and the product marketing side of our desktop business.
It has, and I don't want to say it's been understaffed because we have been performing great, but I think we could get even more out of it than we are. Is that a fair way of putting it?
That's exactly
Yeah
where we're at.
Very good.
Let's continue. We're continuing here in the room, and if you're following us via the web, we'll continue there, too. I put away the iPad with questions. Type in questions on your web app, we'll address them subsequently. I have been allowed today to talk about numbers. When I'm done with that, we'll have André coming back to talk to make a sum-up, we'll address questions once again. That's the way we're going to do that. We have some light lunch for the people here in the room when we are done. I'm going to talk about creating value through profitable growth. That's the same headline as last year. I can tell you, no news there.
What I'm going to talk about is how our priority is profitable growth, how we split that into a few value drivers, and how we address those different value drivers. On the desktop PC, it's about leadership, it's about revenue growth, it's about diversification. It's about margin protection and optimization. On data center, it's more about penetration, getting new customers, of course, stabilizing the margins and operations. You've seen our gross margins be pretty volatile on the data center side in the past. Going further down the profit and loss, we are talking about cost base and its optimization. We are talking about targeting both our IP and especially our R&D investments. We need to be a little cheap and a little tight on the money to make sure we don't overspend, because there's a gazillion opportunities that we can follow.
We need to pick and choose the right ones, of course. It's also about manufacturing. We grew our manufacturing set up last year and built it to enable the increased activity that we have seen, some of that, but where we were hoping a lot more to come. It's about sales. What we do, it's about sales and marketing efficiency. John runs a very lean operation when it comes to both operations, but also on the sales and marketing side. It's about cash flow, of course. It's always an ongoing optimization process, where we go through, on a continued basis, the different line items on the balance sheet to make sure that what happens on the profit and loss actually turns into cash money at the end of the day. Let's take a little bit look at the revenue.
Again, you have seen those numbers before, and André, he stole my thunder by talking about how some quarters are higher than others. This is more a structural overview coming back from Q1 in 2014, and then all the way up to now, where you can see that they do go up and down. Q3 and 4 are typically the highest volumes revenue quarters, and then Q1 and 2 sort of compete about who is the biggest one of those two. This year, with a record $14 million in Q1 2018, is of course a nice beginning to the year. You heard about how we're looking into a strong Q2 also. Very nice beginning of the year. John, he was referring to CAGRs market growth of 5% per year and 7% per year.
I just want to point out that Asetek actually performed at a CAGR, a compound annual growth rate, of 41% in the period of 2014 to 2017. We are definitely eating into our markets, which of course is exactly what the plan is to do. You can also see here that, a little bit of history, we increased our revenues in the 2015 area, or in that time frame, to become at a level where we started making money. The orange line here is the group adjusted EBITDA, which became positive first time in the second quarter of 2015, and have been to various degrees ever since. If we just focus a little bit on this first quarter of 2018. Our total revenues were $13.9 million. Record high Q1, I think we have said that a number of times. We are actually pretty proud about that.
Most of that, of course, comes from the desktop side. 600,000, 660,000 to be precise, is from data center. Yes, that is an increase of 50%, 58% to be specific, over the same quarter the year before. No, we are not happy at all. It is not where that should be. We need to grow that market. Gross margins. What you see on the right-hand side on the top is the year-over-year full year development. The orange line is the one for the group as a total. Because desktop is 95% of our revenue, of course it's the desktop that drives that. The gray line that goes very much up and down is the data center gross margins. They have been fluctuating.
The full year, we talked about also that when we met in February, the gross margin for the full year decreased to 36%, versus 39% almost in 2016. A good chunk of that comes, as we also discussed previously, from currency exchange issues that we saw in 2017. If we look at the lower part, the bottom part of the right-hand side, then it's the gross margins for the last five quarters leading up to Q1 of 2018. I just want to point out that on the desktop side, the dotted gray line, you can actually see a small increase from Q4 of 2017 to Q1 of 2018. That is despite the fact that we have seen exchange rate headwind also in Q1 of 2018.
The initiatives we launched earlier this year and late last year about sales price increases to our customers and fighting backwards in the value chain to make sure that we get the best cost prices possible is actually coming to fruition here in Q1. Just quickly, just so you don't accuse me of missing it, the data center gross margin at 15.5%, obviously, is much, much lower than where we want it to be. We want that. We can see a path to, we need that to be much higher and higher than the desktop margins. Simply because of the low volumes in Q1 of 2018, we have not been able to get any meaningful gross margin out of that segment.
Currency exchange rate, just to reiterate it, 2016, the year to the left, if you see here, on the left-hand side of the screen, you'll see the CNY, how that came down in cost by 5% during 2016. That's nice. That means that our cost of goods goes down by a little bit more than 3%. On the other hand, it came up by 6%, meaning a total cost price increase or a total impact on our gross margins by 3.6%, I believe it was in 2017. We have been extremely impacted by the currency exchange rate and the CNY going that direction. It has stabilized for now. I don't know what's going to happen in the future, but at least it has stabilized during the first quarter of 2018.
However, it's still 3% higher than it was on December 31st, and that's a part of what you see on the gross margins in Q1. On the DKK versus USD, we saw the same picture in principle. It was just much worse, in the sense that the DKK appreciated by 12% during 2017. Since 80% of all our overhead expenses are carried in DKK, of course, that makes for a more expensive operation here in Denmark. That is a structural issue we have going on, selling in USD and buying our services in DKK. The DKK versus USD has also flattened out, it appears, for now. However, it's a little bit difficult to see in the graph here, but it is actually 3% higher than it was on December 31st.
Taking a look at the full picture of the earnings at the EBITDA level, going all the way back to Q1 of 2014. What we can see here is that, yes, we have structural changes, deviation changes in our revenue quarter to quarter. However, when we started selling for real around 2015 and onwards, we also saw the EBITDA margin for the desktop side, which is what pays for the expenses in the rest of the business, go up to a level around 30%. That's the gray line you see on the top on the left-hand side. Yes, it fluctuates, but it balances at around the 30%, which I believe is pretty nice for a component business like ours.
What's also interesting here in my book is that the orange line you see on the left-hand side is our overhead expenses for the entire company, which is paid for by the desktop business. It's fairly flat. You will tell me that, "Peter, $4.2 million," as it says on the right-hand side, "is not flat compared to the $3.5 million or whatever this is on the left-hand side." Keep in mind that the impact from the more expensive USD is impacting us here. Had we had the same exchange rate in Q1 of 2018 as an average in 2017, our expenses in Q1 would have been $350,000 lower.
It does have an impact, and it does underline my argument that we are operating at a higher level of revenue with the same amount of machinery, the same infrastructure, which I think is something we should be proud about. On the right-hand side of this slide here, it is the contribution in terms of $. No news there. The gray bars on top is the desktop contribution, solid contribution, all out, paying well for the efforts we are spending in the data center business. Here, we see that this data center business has been more expensive in the last few quarters. Some of that is due to exchange rate. I'm getting tired of saying exchange rate.
Also, of course, because we have spent quite a lot of resources on the data center business to get the Intel agreement to fruition and get the products out the door. Summing up on income statement, we have talked about most of it, so let me just briefly go through here, coming down from 30,000 feet to just this quarter. $14 million, almost, of combined revenue. A gross margin of 35.7%, which is down from the 38% last year. However, if we look at the gross margin for desktop in Q1, then it is down by not as big as amount as it would have been if we had not been affected by the exchange rates. We do see a positive effect from increasing sales prices and fighting back on cost prices. Operating expenses is going up, again, primarily for exchange rate issues.
We see a lowering overhead expense on the desktop side, and we see more emphasis on the data center side. No news there. It's the same. EBITDA margin of 12.6% versus 10% last year. Again, not a big surprise. We then come to depreciations, that requires a comment. It's increased a lot, and that is simply due to the fact that Maybe not simply, but it's primarily due to the fact that we have built a lot of projects that turns into products that we need to write off as we start selling those products. And we have launched new versions of our desktop products recently, which is a part of why we see an increase in the desktop sales. And those Projects needs to be written off, and we do that over 18 months.
There's a lot of things going on here that have been finalized and that we need to write off. And then we see a share-based compensation line also up to $300,000 between the two segments versus 48 last year. That is primarily driven by the fact that the warrants that we granted last year were granted at an exercise price, which was driven by the share price, which was much higher in May last year than the year before when we granted the previous ones. That is, it's paper money. It's all driven by the Black-Scholes calculation that is coming into effect here. EBIT, earnings before interest and taxes from the two segments, $592,000, 4.3% of revenue. And then from that, we deduct our, what we call our headquarter expenses, which is a good chunk of that is litigation expenses.
We are busy fighting various lawsuits and claims from people out there. That ends us down at a negative 382 for the quarter, which is a diluted EPS, earnings per share, of 4% to the negative versus $0.01 to the positive to the year before. Looking a little bit about cash, at cash generation and usage. Again, going back here three years. The drivers, well, we had a financing round back in 2015 that brought us $12 million. Apart from that, cash generation and use is primarily driven by EBITDA from the desktop business and the data center business. Desktop a positive, data center is a consumption. Both during 2016 and 2017, the positive contribution from desktop has been around $15 million, $16 million. Then we are consuming in 2016, $5.1 million, and in 2017, $7.3 million to get the data center business up and running.
Last year, we also paid out a dividend of $2.9 million. That altogether, all these things combined, ends out with a positive cash flow in 2017 of a million and a half, ending out at $18.4 million. We have then, here in Q1 of 2018, spent a little bit due to primarily investments and changes in working capital, meaning that our cash balance is now $17.7 million. Cash conversion, if we start at the bottom over here, of course, this is the important part. How do we turn our profit and loss into real cash and do it as fast as possible? The cash conversion in number of days in 2017 was four days. We have a soft target. We would like to be able to collect from our customers at the same level of days as we pay our vendors. That's the zero. That's our soft target.
It was four days in 2017. We believe that's pretty okay. When we measure them by quarter, the calculation gets a little bit different, that means that at the end of Q1 in 2018 was up at 11 days instead, versus 29 days in the same quarter of 2017. A low number is a good number here. Preferably a negative number, because that means that we are getting paid from our customers earlier than we are paying our vendors. We've had that once, just like Dell. The primary drivers in all this, the primary drivers of our balance sheet is our accounts receivables and our accounts payables. No news there. It's pretty much as it used to be. We have generally payment terms with our customers of 60 days, they generally pay very much to those payment terms.
We then turn around and pay our vendors in 60-plus days. Balance sheet, no news there either. We have a solid balance sheet. We have a solid cash position. We have low, almost no interest-bearing debt, that means that we're ending out with a, what we call a lean balance sheet and a flexible balance sheet, allowing us to be strong when we talk to new customers, new vendors, et cetera, appear to have a strong and agile balance sheet. Financial year, fiscal year 2018 financial outlook, talking about the same topics as we did in the beginning of the presentation. It's about profitable growth. When we look at revenue, the desktop segment, we're expecting growth between 10% and 20%. You heard John talk about the assumptions behind those numbers. On the data center segment, we expect the development to be flat.
You also heard both John and André talk about the assumptions behind that number. Gross margins, we expect on the desktop side to be stable, probably increasing ever so slightly on the gross margins. The data center side of things, on margins, we simply just need to get more scale and get the margins up that way. There's no other way than getting to scale. Capital allocation, primarily overheads, R&D, DKK 3 million to DKK 5 million. It was a little bit higher in 2017 than it was in 2016. We probably going to stay at the same level this year. Headcount is more or less constant, and that means, of course, that since two-thirds of our overheads is related to headcounts, if headcounts is fairly stable, then our overhead is also fairly stable. Then we spent CapEx between DKK 2 million and DKK 3 million.
It was a little higher in 2017 because we built the capabilities that we have here in the facilities for manufacturing, et cetera. Also some of the
Tools, the molds, and stuff we use for these new products that I talked about on the desktop side, the new generation, were fairly expensive and were invested in last year. That drove that up. Of course, what we want to do is to maintain a strong balance sheet like the one I just showed there before. It's important for us to have a strong cash position to withstand problems that may come our way, but also to appear strong towards our customers and suppliers. Shareholder return, or the Prime Directive, as they call it in "Star Trek," of course, it's our mission in life to have the share appreciate and create enterprise value. Last year, we paid a dividend. We have decided not to do this this year after carefully considering our various options.
Instead, we are considering to make some kind of a form of a shape of a share buyback program to potentially send back money to the investors. You can hear from my phrasing here that that is very much in its infancy still. Is it on? Okay. I think we said it enough times, everything in the past slides that you'll remember it. I think there's some other things I would like to emphasize here. Of course, it's annoying to look at the share price right now is down 8%. I'm losing money also by that, of course that's not fun. However, my job is not to drive the share price up. My job is to do the best for the company, that's what I'm doing. That's what I'm trying to do, then hopefully, that will be reflected in the share price at some point.
Just to keep that clear. The reason I'm saying it is it's a huge disappointment that the Intel revenue is where it is or where it looks to be. Keep in mind, as I told you, I'm trying to be transparent. I'm telling you what I see. I'm trying not to judge it. I'm just trying to tell you what I see. It could change in any direction. If you look at this from a company perspective, it's actually a pretty big achievement that we've been able to actually get a customer like Intel. Of course, what is it worth if they don't end up buying what we would have liked to? I don't really think it's a day-to-day that's sad and it's bad. We have a booming desktop business. We are making more money than average companies in our space.
The outlook for the desktop business is really strong. We have every opportunity to go and capture the data center market, which I'm sure we will. Yes, it takes a little bit longer, I think it's also important that we could have been standing here today and said, "The Intel collaboration fell apart because we were not up for the task. We had quality problems. Our engineers were not smart enough, et cetera." Then it's a closed door. We now have an open door with Intel, which is one of the strongest brands in the industry. I think just reflect a little bit over that. I think that's worth something. I would welcome any questions that you have. As I said, we'll try to be as transparent as we can.
I would welcome you to, for those in the room, of course, to take a look at our new facilities and get an impression of what it is. What you can see here on the right is a rack, and there's some server nodes, and what you can see here is a desktop PC. Now that you're in the house, I would invite you also to take a look in our R&D department and see what our people are actually doing and what it requires to get to products that we have right here. I think, with that being said, let's open up the floor again.
Yep. Mr. Anderson from SEB.
Hello again. You talk about the price increases. Perhaps, with the Q2 revenue expectations, could you shed some light on where gross margins then might end up? Secondly, related to that, the price increases that you mentioned in your Q1 report, how much is actually coming through there? How much have you actually raising prices? Finally, on the overheads. What you're saying is, Peter, that the OPEX that we saw, which I calculate around DKK 18 million last year, that should be stable from here in 2018. If we're going to see the revenue increase that we all hope for, that number's not going to rise to any significant. Is that correct?
Just to make sure we're not playing tricks here, any of us. If you convert it into DKK, yes, it should be flat. That's a correct assumption. In terms of price increases, again, with the currency exchange in mind, I cannot say where that is ending up, but if you do not factor that in, for sure you can see the effect of our price increases. I think you already can. As I told you guys in Q4 that we were doing it, and we were, but of course, we had to burn through inventory and now we are selling at the new prices. For sure it will have an effect. In terms of the actual rate, I think you're perhaps better positioned to answer that, but there is probably an average of 3%-5% or something like that, I would say.
Yeah. I don't know if all that came through, in terms of the margin for Q2, I would model it to be pretty much as it is today. In Q1, sorry. Yes, we have seen some continued price increases at the customers, but it's minute numbers. I would model it as Q1. Overheads, you spoke about overheads. No, in absolute currency dollars, currency kroner for that matter, we are continuing at the level we were at earlier. No, nothing will explode just by more revenue coming in.
Thank you.
Any others? Should I liberate you of the microphone? Thank you. I do not see any new questions on my magic thingy here, which is actually strange because it's been here since we left for the break. It is updating, right? It is updating. Yeah. Okay. Cool.
All right.
Any other questions? That concludes the presentation. Thank you for coming. Thank you for your interest in Asetek.
Thank you.