Raspberry Pi Holdings plc

Thank you all. Thank you all for coming. Thank you to the people who are joining us online. Yeah, this is our interim results for our very, very first interim results announcement. We're just going to take a moment to kind of run you through the business. I guess quite a lot of you know what it is we do. Some of the highlights from the first half. Richard's got the numbers. And then just a little bit of a discussion of our progress against some of the strategic goals that we set ourselves at the time of the IPO. I think we're going to try and go through this relatively quickly to leave plenty of time for questions at the end. So do I need my clicker over here? I'll hop up here. So here we go. Click. OK, just briefly who we are. I'm Eben. I think many of you know me. I'm a silicon engineer by background, and I've been running Raspberry Pi. I founded the Raspberry Pi Foundation back in 2008, and I've been running Raspberry Pi since we pulled it out from under the foundation at the town end of 2012.

And I'm joined by... Yeah.

The numbers go up pretty quickly, don't they? Yeah, there we are. Cool. So a very quick recap of who we are. We build compute platforms. By compute platforms, what do we mean? We mean hardware. We mean the software that runs on the hardware. And we mean all of the collateral that sits around that hardware and that software. to allow users to get value out of the systems that we build. We sell to educators and enthusiasts. That was probably the original market back in 2012. But increasingly these days, our dominant business is selling to industrial and embedded customers who are using our products in industrial contexts and outsourcing the intelligence components of their larger products to us. Our products are designed and built in the UK. We do almost all of our R&D in Cambridge. We do almost all of our manufacturing with our partners at Sony UK Tech in South Wales. And over the 12 years since the launch of the first Raspberry Pi on the 29th of February 2012, we've sold a little over 60 million units. And of those 60 million units, roughly 90% of them have been exported outside the UK. So some highlights from our first half, and we'll go into a little bit more detail in a while. Obviously, the IPO was the standout event of the first half. It was a $185 million sell-down for our larger shareholder, the Raspberry Pi Foundation, setting them up to do the good work they do at scale for the next decade. And a $40 million primary for us, which we're using to fund our technology development roadmap and to increase the robustness of our supply chain position. Highlights in terms of the market, we now believe we've completed our recovery from the shortages which were associated with the aftermath of the COVID-19 pandemic. There is still turbulence, but those shortages which were hurting our business until about 18 months ago, now largely abated. We're seeing good progress. As I say, there is turbulence out there in the world, but we are seeing good progress now towards the normalizing of channel inventory downstream of us. And we haven't let the IPO or the turbulence associated with COVID stop us from innovating. It's been a good half year for us. We launched our first AI, our first first-party AI product. We launched our first cloud software product. And we completed the ramp of our second-generation microcontroller platform, RP2350. So, and at the same time, made solid commercial progress. This was the, we launched Raspberry Pi 5, we announced Raspberry Pi 5 almost exactly a year ago to the day. The first half of this year represented the first moment when Raspberry Pi 5 was available in channel, ex-stock in channel. We've seen a good transition from Raspberry Pi 4 to Raspberry Pi 5 as the most popular product among our enthusiast customers. We've made progress on our goals to expand our ASP and increase our accessory attach. And we've seen a return to growth. We've been able to allow ourselves a return to growth in our approved reseller base. The way we reach our customers generally is through a network of a little over 100 resellers around the world. We paused the growth of that reseller base during the supply chain shortages, but we have now been able to get back to that. So with that, I'm going to hand over to Richard to introduce the numbers.

Good morning. So some financial highlights. I'm just going to run through it and then a little bit more detail to follow on. We finished the first half having sold 3.66 million in what felt like a strong first half, though admittedly against the supply constraint 2023, we were still coming through the end of the semiconductor shortage in that first part of 23. I think we successfully, in this period, navigated through that supply chain shortages and the ripples that came from that. That unit number was probably a bit a touch lighter than we were expectations. But against that, we've had better than expected profit per board, $8.30 compared to $7.70 in the equivalent period a year ago. That gave for gross profit of $34.2 million, which was up 47%. And that led to an adjusted EBITDA of 20.9 million. That's up 55%. As you can see, our costs mathematically didn't increase at the same pace as that profit. And then adjusted operating profit of 15.7. We had a gross margin of 24%, and I'll talk a little bit more later about that. So in terms of units by channel, really our two core channels, the direct channel, which is where we sell, we report that revenue as our revenue. And we're selling to resellers and also within their OEMs. But H123 was actually, we managed that one extremely well. So I think the volumes there were good. Farnell has really come out more into this year. And then we've also got the switch of selling Pi5, which initially we're selling the In this period we were selling the high cost 4GB and 8GB variants, which I think as we've explained is very much planned. That's where we rely on Farnell with their balance sheet strength to launch those initial products. So we saw, as you can see, some growth in our direct units. And then within the licensee, really a quite significant growth as Farnell came through. And I think in the past, we've guided to being 20% to 30% of our units we'd expect to see going through our licensee partners. In this period, it was about 35%. That's very much the effect of the Pi 5 launch and the growth in volumes there driving back to a slightly higher level. And I think we'll see that start to come down over the second half. Revenue breakdown by category. Half jokingly, it's a graph that goes up and to the right. Publishing is a small part of the business. Ironically, it's the first thing on there. Just those are the magazines. Royalty income. So this is the license we get from Farnell for those boards. The boards grew by nearly three times. Royalty increased by, well, exactly three times. and a bit, and that's really an effect of being slightly better royalty per board on the basis of the Pi 5 attracts a little bit more. Components, those are the memory and logic chips that we sell to Sony who build them into the Farnell boards. We report that as a separate line. We increased the volume of that. Obviously, as Farnell's volume increased, we sold more memory to them. But we also, in the period really to manage that initial introduction of the new chip that goes into the Pi 5, we sold directly to Sony, whereas in the past, most chips go straight to Sony from the manufacturer. So there is a bit of variability in that. It'll depend on our commercial activity, which is very much the reason why we encourage people to look at the gross profit as a key metric rather than revenue. But I appreciate everybody kind of wants to see that number. In terms of products, so this is where we're selling the finished boards that we sell. Also accessories, which had a good performance in the half. And we'll talk a little bit about that. And that grew to 89 million through the period. A bit more detailed. I hope it's legible on the little books you've got. Otherwise, it's available at all good booksellers. Revenue was up overall 61%. Gross profit, which we touched on, was up 47%, which was $11 million. With unit sales, the board's up 31, so that was driving it. The average selling price of those boards increased by about 28%. We were selling more Pi 5s in this period compared to, say, things like Pi 3s a year ago, which again, there's another step up in price. And I guess also the Pi 5s we were selling were the ones with more memory, so they were the $60, $80 boards was also coming through. Gross profit per unit increased 8% to $8.30 from $7.70 a year ago. As you can see, that's obviously a lower growth rate than the ASP. Pi 5 is a lower margin. It does make less margin than the Pi 4. That's partly because of this initial cost of the first tranche of Broadcom chips that we use. But also, those chips are more expensive overall. So there is a slight reduction in the margin per board for that reason. Gross profit per unit, I think the other sort of nice part of the year, another feeling why the first half was good. So the unit from accessories, our profit came up 38%. The total gross profit came up 74% per unit, so per board sold. It also rose 38% to about $1.10. And that was a total of about 4.1 million of profit. Accessories really sold well in the half. What was that? Partly Pi 5 has clearly attracted a new tranche of accessories such as power supplies and cases and cooling add-ons. It's also actually the price of the panels that go into our displays and screens has come back to normal, has improved for us, and so we've seen more profit coming through from there as well. So overall, good performance on accessories, which I think we sort of feel should continue into the second half. In terms of overall administrative costs, and we split those between two areas, the cost of R&D cost, as we call it, which is essentially the engineering team the part of the engineering team that we don't capitalize. We don't capitalize on software because it's generally free release across the market. And inevitably, a team also has other pieces of downtime that we wouldn't attribute to any particular product. So that increased about 45%. I think we added about 36% more engineers into the team. They were generally skewed more towards the hardware and semiconductor side. So in most areas, they would expect to be things that we've capitalized as they're working on, whether it's microcontrollers or improvements to boards. So overall, that grew by about 45%. Administrative costs, that's the rest of us, the lotus eaters like me. That grew 34%, so about 9.1 million. That was There was a general increase in salaries of between 5% and 10% across the organization, a few more heads, and then really just the costs of becoming a bigger business, the costs of more audit fees, bless it, the cost of an interim audit review, and just generally we've also been working on systems, which has added some cost into the period. So overall, that grew by 34%. I think the good message there, it still grew by less than our gross profit. which therefore means that our adjusted EBITDA increased by 55% to 20.9 million. Depreciation amortization is up. It nearly doubled as we started to amortize the Raspberry Pi 5 and the RP1 chip that goes into it. That total package, if you will, is about 20 million of capex. It's been building over the last few years. The RP1, as I think we've explained to people, is enormously valuable as a way to embed one of our better moats. In fact, it's probably a wall that it basically gives a Raspberry Pi its Raspberry Pi-ness and it's our own silicon. So the ability to clone it is extremely challenging now, which is nice to see. Again, we started to amortize it once it was released. Obviously nothing in the first half of 23 and then into 24. So after just an operating profit, up by about 44%. Effective tax rate, if you take our profit before tax, divide it into the tax charge, it would come out at 29%. I point out within there, we've got about 2.1 million of non-deductible IPO costs. So at an underlying level, once you take out those sort of one-offs and adjust the tax, it was about 25%, which I'm sure is what David Tweedy and all those people intended when they came up with the deferred tax standards and all the other horrors of the accounting canon. Adjusted EPS grew. Basic EPS went down, as we said. There were some significant, unsurprising one-offs in the first half, be they share scheme costs or be they the costs of the IPO. Turning to a balance sheet... At the top, intangibles. Intangibles grew from the end of the year at 58 to about 65 million. We spent about 10 million or so on R&D, capitalised R&D, very much in line with the guidance we gave. I think we guided to about 20 million for the year, so 10 in the first half. Tangible assets as small increases as we continue to fit out a warehouse. Inventory grew. It's 146 million at the end of June. That compares to about 108 million at December 23. That's an increase of about 35 million. There's probably a few things to pick out there. Finished goods increased by about 20 million. That increase is probably split broadly into two categories. There's about 10 million, which I think is fair to say is the cost of having the right amount of inventory for compute module and Pi 5, which at the end of last year, we were either only just ramping up or was still on back order, whereas we're now into a more steady state. The other 10 is undoubtedly we have more inventory than we would choose on things like Pi 5s, sorry, Pi 3s, where it has been slower in the half. I think we've flagged and it's sort of fairly well acknowledged industrial sector has been some overstock. We've certainly seen it in that space where the sales that we're making are a bit slower. And so that has built up into inventory. In terms of components, raw materials that we call it, that grew by about 15 million. Our memory that we're holding, we had 35 million of it at the end of last year. That's probably about 5 million higher or so now. We continue to see some quite interesting opportunities to buy memory. Despite market prices being up 50%, the sort of spot price you see quoted, Evan and Mike, our commercial director, have been still receiving sort of invitations to buy memory, particularly curiously at quarter ends when somebody's bonus scheme is probably looking a bit thin. We've seen those opportunities to buy memory into next year. So we're probably covered on that sort of key DDR4 memory through into probably the end of the first half of next year. pretty much at the rates that we saw in 2023, which is a nice situation. But we will continue to do that sort of opportunistic purchase of memory as it gives us such a sort of security of confidence in the profitability into the future. And the other part of it is we've been buying some or bringing into inventory logic chips that have come through, particularly, I suppose, from Broadcom that go into our products. And that stepped up and there's probably about 10 million of those now, which is all part of our strategy, really, of making sure that we have a sort of secure source of those chips. Because for us, I think that memory of a year and a half ago of just not being able to get the products out we wanted because of that shortage. Other ways, trade payables really is probably the other part on here that I just pick out. We have seen with some of these purchases of inventory on the component side, people offering us quite good payment terms as well. So you've seen a growth in that trade and other payables to 71 million compared to a year ago, compared to the end of the year. That number at the end of the year had, we'd bought most of that 35 million of memory in December. So that was literally there at the end of December. We paid for it in January, February. So it went out pretty quickly after that. Some of those payables that we're now seeing actually do have a bit longer periods on them, which is nice. So we finished with cash of 40 million. Talk a little bit about the cash flow over the page, which was after IPO proceeds of 31.8, which is basically the 40 million that we raised minus the fees that have gone through share premium, the costs that have gone through share premium. Very quickly on the cash flow, I think I've covered quite a lot of these areas. Started this year with 42 million EBITDA in the period, the 21. We have increased inventory by 37 million. CapEx, as I said, is about 10 million, 11 million in total in the period, IPO proceeds Paid some tax, which is a bit of a novel experience for us, but if we make profits, it's probably what we have to do. And then other costs, 2.4 million, so other sort of miscellaneous expenditure, including, I think, some of the other costs relating to the IPO that we've shown as an exception. Leaving us with about 40.4 million at the end of the half. So overall, I think it was a good H1. Very pleased when we got to the end of June. Slightly lower units, as we said, than we would have liked. Profit per unit, though, was better than expected as a fact of mix, I think. We've got through nearly all of that expensive or more expensive Broadcom chip that goes into the Pi 5. So we're nearly through that sort of $5 increment that we talked about at the time of the IPO roadshow. And we've seen as a pendant to that, obviously, Pi 5, about 1.1 million units, as Evan mentioned, sold in the half. So we're basically seeing that Pi 5 launch work through. So a good half overall. I'm just going to touch briefly on how that sort of maps against what we've sort of said are our strategic priorities. Grow unit sales. So in the half, we were up about 31%. So I think we... That was good. Unit profit, obviously we got to $8.30, up from $7.70, so unit profit increased in line with that. Profit participation, so what share of the total profit between the manufacturing price and the end price to a customer do we retain well that came down a little bit because of the switch or the predominance of the sales through the licensee channel in that second half and that's something sorry first half that's something else we'll continue to work on so I think overall a lot of the things that we said we would do we've continued to do we'll continue to work on those other areas and I will now hand back to Evan to talk about what that means as we go forwards Thank you.

Let's have a quick look here. Good. OK, so a little more detail on those first half highlights. Raspberry Pi 5, as I say, this is the half in which we We ramped to X stock availability of this product. Over a million units sold in the first half. This is a pace of transition which is, we don't do major product transitions very often, so we don't have an enormous data set of what they look like, but this is a pace of transition which is fairly consistent with what we saw in 2019, 2020 as we transitioned people, as we transitioned certainly our enthusiast customers from the Raspberry Pi 3 to the Raspberry Pi 4. platform. Interesting about this, of course, it is a vastly more powerful product than its predecessor. It's roughly three times as powerful, three times the performance of Raspberry Pi 4, about 100, to put that in context, that's about 150 times the performance of the $35 Raspberry Pi. one that we launched back in 2012. The interesting thing about it though, although it looks very similar to its predecessor and it is a more powerful device, is it also embodies our latest thinking about what good manufacturing technique looks like. We introduced a number of improvements in the design and manufacture of the product, in particular the way that the larger connectors, the through-hole connectors here are mounted into the board, and the way that the individual boards are singulated. Those of you who've been to the factory will have seen that Raspberry Pis are manufactured in panels of nine. We've changed the way that we singulate, this process of singulation by which we break Raspberry Pis out of those panels. This was largely intended to deliver manufacturing cost improvements, but it's also, it turns out, we'd expected this, but it's been surprising to see how effective this has been. It's also delivered quality improvements as well. We're seeing roughly a 50% lower early life failure rate in Raspberry Pi 5 products than we did in previous generations of the products. And the expectation is we're now going to take some of those insights that we gained in the manufacture of Raspberry Pi 5 and roll them back, certainly to Raspberry Pi 4, maybe not to the earlier products, maybe not to Raspberry Pis 1, 2 and 3 as the volumes of those continue to taper away. But we do feel that we can realise some cost savings and some quality improvements, certainly to that previous generation of product. I think I see, apart from the IPO, I see 2024 as really being a year of new product launches. Just to pick three of these, pick three of the most notable ones, one of which, the first one actually, launched the week before the IPO. People have been doing AI with Raspberry Pi. AI is very buzzy and very zeitgeisty at the moment, but people have been doing things that you would recognize as AI and machine learning applications with Raspberry Pi since 2012. The AI kit, which we launched in June, though, it's a watershed moment for us because it's the first time that we've built a product ourselves and put our own logo on a device which is designed specifically for AI acceleration. This is a partnership with our friends at Halo. It bundles their 13 trillion operation per second accelerator board with our our M2 hat, which connects it to a Raspberry Pi 5. We've seen very, very robust demand. This is one of these products you never know when you launch an accessory, whether it's going to catch on fire or whether it's going to be a niche product. This is one where every time we've dialed up the production rate of this product, we've seen the market draw them from us. Obviously, like all of our products, in the early days it tends to sell to enthusiasts, to hobbyists, but we're confident that we're going to see a transition of this product and its successors into our industrial customer base over time. And I should say it is the first of many, first of several AI-specific products that you'll see from us over the next little while. The second one, Raspberry Pi Connect. For us at Raspberry Pi, software, it's a very important part when we talk about compute platforms. We're not just thinking about things you can touch, things made of atoms that you can put your fingers on. We're also talking about software, things made of bits. But for us, software at Raspberry Pi, it's always been a thing that's run on the device. And it's always been a cost center for us, not a profit center. It's never been something that we've attempted to go out and sell separately from physical pieces of hardware. It's been an enabler for the hardware and an enabler for our customers. So Raspberry My Connect broadens our notion, I guess, of what we mean when we say computer platform away from simply pieces of physical hardware and the software that runs on them to things that run in the cloud. Now, this is a solution that provides our enthusiast customers and our industrial customers with secure remote access to Raspberry Pi devices in the field. It tries to embody that kind of just works mindset, that just works philosophy that we try to have all our products embody this. You can put your Raspberry Pi anywhere in the world, and you can then connect to it from anywhere in the world and control it very, very easily. So it's the first time we've built a compute platform that runs out in the cloud. It will also in due course, this is in public beta at the moment, it will also in due course be the first time we have charged money for something that isn't made of atoms. So it'll be the first time that we've charged our industrial users of this platform money for that remote access service. And we've seen a lot of interest in the sort of the upcoming, I guess, enterprise upgrades. of the software platform. As I say, public beta, strong interest, 50,000 users since we soft launched it in May, over 100 expressions of interest in upgrading from that free tier to the enterprise tier when that becomes available over the next few months. And finally, we haven't stopped innovating in silicon. Raspberry Pi Pico 2, which we launched. That's the second half product. We launched that in August. But the ramp of the underlying product, the ramp of RP2350, the underlying semiconductor product that enables it, that occurred in the first half. So this is our second-generation microcontroller-based SPC. It's built, as I say, on RP 2350, second-generation microcontroller platform, which provides more performance, improved interfacing, and some of the security features that our customers have told us they need if they're going to start deploying our semiconductor products, our microcontroller products, in preference to legacy incumbents, and almost perfect compatibility. with software which was developed for RP2040, its predecessor. And you'll see more products from us over the next little while with that microcontroller in. So what do our go-to-market priorities look like in the second half of 2025? Just picking three, that gradual shift towards the direct-to-OEM channel model, that provides us with improved unit economics. It provides us with improved touch on our OEM customers and improved both more money and a better ability to understand what we need to change about our products in order to succeed in the future. We have found, interestingly, that the credibility that's come with being a public company has been opening doors for us. It's enabled us to engage much more easily with new prospects at a more senior level in those organizations. As I said earlier, we are resuming the expansion of our approved reseller network, having put that on hold during the shortages. Really, the target there, find those geographies which are underserved, find verticals which are underserved, find verticals which have vertical-specific reseller partners who can really go and drive our products in there. And finally, we're continuing our investment in both in-house and third-party sales resource, again, about underserved geographies and underserved verticals. And there's a particular focus there on leveraging the demand creation capabilities that are inside our major distributor partners. Many of those organizations have hundreds of application support engineers who we can educate to go out and sell our products into industrial customers. What does that look like when we get it right? Well, this is probably the standout recent example of a scale deployment of Raspberry Pi in the UK. FIDS, Flight Information Display Systems. That's the word for all of the big screens that you see when you go to Heathrow. It tells you how late your plane is. The solution that if you were to go to, if today you were to go to Terminal 5, and I did this the other day, and you can see how old they are. All of that infrastructure in the Heathrow estate is very old, sort of roughly a decade old, approaching end of life. Platforms which are no longer available, platforms which are failing. There was a challenge to modernize the infrastructure the signage across the entire estate while achieving best price performance. And so in partnership with Sharpa and NEC, we went out and won this business. Every display you see now in terminals three and four, certainly ground side, every terminal you see now in terminal three and four, It's a sharp NEC display powered by a Raspberry Pi Compute Module 4. How do we win this business? Well, we offered a solution which was flexible, adaptable, secure, and critically manageable. We offered a low acquisition cost. The upfront cost of the hardware was low. We offer a low running cost in that this is extremely reliable hardware that does not need to be replaced frequently and also consumes little power. And then we offered longevity. Compute Module 4, which we've shipped into this opportunity, will be available. We tell people that will be available at least through 2034. One of the realizations we've had as we've increased our participation in industrial embedded cells is how important it is for people not to feel they're going to wake up one morning and find that the thing that they've invested in is no longer available. So longevity was a very important part of winning this business. Obviously, this opens the door to new opportunities. There are further opportunities within the Heathrow estate, both to complete the rollout of this to Terminal 2 and Terminal 5. There's an entirely separate opportunity around BIDS, the baggage information display system, so the things you see inbound, the thing that I spent an hour and a half staring at on my way back from Shanghai on Sunday night. There's an opportunity to go win those. And of course, Heathrow, it is a flagship design win. It's an opportunity to go and then sell similar solutions into other airports elsewhere in the UK and overseas. A little bit about people. Obviously, I've said before, our people and the culture of our organization are the deepest moat that protects Raspberry Pi, our main source of long-term differentiation. We've continued to grow the team. Our number of employees up 21% year on year to 115. So that's an average across the first half of 24 versus an average across the first half of 23. An earlier version of this slide said our number of average employees was up 21%. I happen to say our number of average employees is up 0%, so we're continuing with that trend. We continue to invest in the engineering team. Engineers account for over 50% of our employees. All of our employees are critical to the success of the organization, but we do track that fraction of the money that we spend that's turning directly into new products that will we can offer our customers. We have a high rate of stock ownership. I think before the IPO, we were distinctive in having had a very, very broad employee share ownership program. We are very gratified that many of our employees have chosen to retain much or all of their holding through the IPO. We've seen very nearly 100% retention rate in engineering after the IPO. That was a very, very pleasant surprise. I mean, I think we had probably budgeted that we might lose one or two people. We've seen 100% retention in engineering. Very nearly 100% retention across the entire organization. What does the world look like at the moment? I think we've been good historically at hiring mid to senior people. So we've been good at hiring people in the middle of their careers. What's changing for us at the moment? We think we finally cracked how we hire graduates. So we have an absolutely outstanding internship program. We're getting better at converting those interns into graduate employees. One interesting aspect of that, we all know that we work in an industry that has a diversity problem. The stream that we get there has a very different and more balanced, in all respects, has a very different and more balanced diversity mix from the diversity mix that we get when we do senior hiring. And that's just because we're measuring the future rather than measuring the past when we hire a 21, 22-year-old. So a little bit on outlook. Three things to, I think probably three things to focus on. One, like a lot of participants in our industry, we do continue to see elevated levels of channel and customer inventory. They are continuing to normalize. They normalized over, they've been normalizing for the last couple of quarters. We do expect that trend, that normalization to be largely complete by the end of the year. Volumes in the second half, we expect to be higher than volumes in the first half. As I say, this is a year of product launches. So that growth in volumes in the second half driven by products that we've already launched, particularly Raspberry Pi Pico 2 and the 2GB variant of Raspberry Pi 5, both of which we launched in August, and products that we expect to launch over the next few months. Those product launches do tend to be towards the low end of our price performance range. And so where we saw those very strong new decades in the first half, we expect to see those normalised somewhat in the second half. I think at the time of the IPO, we indicated a second half waiting to our profitability. I think having had a stronger first half than perhaps we had anticipated, we're now no longer, we're no longer guiding towards that imbalance across the year. And the overall guidance, our expectations for the full year remain unchanged. So thank you very much.