Q2 2026 Raspberry Pi Holdings PLC Earnings Call

September 24, 2026

Operator: Raspberry Pi Holdings PLC interim results 2026 investor presentation. Throughout this recorded presentation, investors will be in listen only mode. Questions are encouraged, and they can be submitted at any time using the Q&A tab situated on the right-hand corner of your screen. Simply type in your questions and press send. Before we begin, I would like to submit the following poll. I would now like to hand you over to CEO, Eben Upton. Good afternoon.

Operator: Raspberry Pi Holdings PLC interim results 2026 investor presentation. Throughout this recorded presentation, investors will be in listen-only mode. Questions are encouraged, and they can be submitted at any time using the Q&A tab situated on the right-hand corner of your screen. Simply type in your questions and press send. Before we begin, I would like to submit the following poll. I would now like to hand you over to CEO, Eben Upton. Good afternoon.

Speaker #1: Pi Holdings PLC. In terms of the 2026 investor presentation, throughout this recorded presentation, investors will be in listen-only mode. Questions are encouraged, and they can be submitted at any time using the Q&A tab situated on the right-hand corner of your screen.

Speaker #1: Simply type in your questions and press send. Before we begin, I would like to submit the following poll, and I would now like to hand you over to CEO Evan Upton.

Speaker #1: Good afternoon.

Speaker #2: Good afternoon, and welcome, everybody. Welcome to our 2026 interim results. I'm Evan, joined today by my colleague Richard. We're going to take you through the first half of financial results for the business, give you some updates on progress against our strategy, and I think first, perhaps, we could just take a quick look at some of the highlights of the half.

Eben Upton: Good afternoon and welcome everybody. Welcome to our 2026 interim results. I'm Eben, joined today by my colleague, Richard. We're going to take you through the H1 financial results of the business, give you some updates on progress against our strategy. I think first perhaps we could just take a quick look at some of the highlights of the half. I mean, quite simply, this was an exceptional H1 for us. Saw record revenue, record profitability in the business. We shipped over 4 million units. We shipped 4.2 million Raspberry Pis in the half, with our customer order backlog doubling to 2.6 million units. All of this happened despite an incredibly challenging supply chain environment. Our diversified DRAM supplier base helped us weather the worst of that, helped us keep our products in production and in availability through the half. Despite these challenges, we kept innovating.

Speaker #2: I'll be quite simple: this was an exceptional first half for us. We saw record revenue and record profitability in the business. We shipped over 4 million units; we shipped 4.2 million Raspberry Pis in the half, with our customer order backlog doubling to 2.6 million units.

Eben Upton: I mean, quite simply, this was an exceptional H1 for us. Saw record revenue, record profitability in the business. We shipped over four million units. We shipped 4.2 million Raspberry Pis in the half, with our customer order backlog doubling to 2.6 million units. All of this happened despite an incredibly challenging supply chain environment. Our diversified DRAM supplier base helped us weather the worst of that, helped us keep our products in production and in availability through the half. Despite these challenges, we kept innovating.

Speaker #2: And all of this happened despite an incredibly challenging supply chain environment. Our diversified DRAM supplier base helped us weather the worst of that, and helped us keep our products in production and available throughout the half.

Speaker #2: Despite these challenges, we kept innovating. The half saw five new product and platform launches, improving support for AI workloads on Raspberry Pi 5, helping our OEM users through the memory crisis by giving them more memory density options on Raspberry Pi 4, and streamlining IoT deployments through improvements to our Raspberry Pi Connect platform.

Eben Upton: The half saw five new product and platform launches, improving support for AI workloads on Raspberry Pi 5, helping our OEM users through the memory crisis by giving them more memory density options on Raspberry Pi 4, and streamlining IoT deployments through improvements to our Raspberry Pi Connect platform. We kept investing in our OEM customers too, making targeted hires into our commercial team, and our board-to-board program continued to deliver traction in the smart home and defense sectors, with a number of meetings with members of the C-suites of UK listed industrials. We kept investing in broader operational capacity too. We hired Tim Mamtora in March as our first Chief Operating Officer. He and his team are already building the repeatable processes that will allow us to continue to scale our business without sacrificing the engineering-led culture, which makes Raspberry Pi, Raspberry Pi.

Speaker #2: We kept investing in our OEM customers too. Making targeted hires into our commercial team, and our board-to-board program continued to deliver traction in the smart home and defense sectors, with a number of meetings with C-suite members of the C-suites of UK listed industrials.

Speaker #2: We kept investing in broader operational capacity, too. We hired Tim Mantora in March as our first Chief Operating Officer. He and his team are already building the reputable processes that will allow us to continue to scale our business without sacrificing the engineering-led culture that makes Raspberry Pi, Raspberry Pi.

Eben Upton: We kept investing in broader operational capacity too. We hired Tim Mamtora in March as our first Chief Operating Officer. He and his team are already building the repeatable processes that will allow us to continue to scale our business without sacrificing the engineering-led culture, which makes Raspberry Pi, Raspberry Pi. Together, these developments give us confidence in our outlook for the rest of the year. We're expecting higher volumes in the H2 than in the H1, and we're expecting full year adjusted EBITDA to be above market consensus. Now I'll hand over to Richard for the financials.

Speaker #2: And together, these developments give us confidence in our outlook for the rest of the year. We're expecting higher volumes in the second half than in the first half, and we're expecting full-year adjusted EBITDA to be above market consensus.

Eben Upton: Together, these developments give us confidence in our outlook for the rest of the year. We're expecting higher volumes in the H2 than in the H1, and we're expecting full year adjusted EBITDA to be above market consensus. Now I'll hand over to Richard for the financials.

Speaker #2: So now I'll hand over to Richard for the financials.

Richard Boult: Thank you. If I can control this. Great. The things Evan said, this was, I think across almost all metrics, a very good H1. We had a strong improvement in unit volume. It continued to grow half on half, as you can see from the chart. It has really progressed from, actually the end of 2024, each half we have consistently grown, finishing at 4.2 million units. It was particularly good to see the demand was in Compute Modules, Raspberry Pi 4, Raspberry Pi 5, our higher margin, our higher value boards, our latest boards. That indeed, that demand was strong that it left us with a backlog of units at the end of June 2026, compared to 600,000 at the end of 2025, to 2.6 million now. Over the period, we have had to increase prices because of memory increases.

Speaker #3: Thank you. If I can control this.

Speaker #4: Great. So, I think, as Evan said, this was, I think, across almost all metrics, a very good half. We had a strong improvement in unit volume.

Speaker #4: It continued to grow half on half, as you can see from the chart. It's really progressed from actually the end of 2024, each half we've consistently grown, finishing at 4.2 million units.

Speaker #4: It was particularly good to see the demand was in compute modules. Raspberry Pi 4, Raspberry Pi 5 are higher margin, are higher value boards, are latest boards.

Richard Boult: It was particularly good to see the demand was in Compute Modules, Raspberry Pi 4, Raspberry Pi 5, our higher margin, our higher value boards, our latest boards. That indeed, that demand was strong that it left us with a backlog of units at the end of June 2026, compared to 600,000 at the end of 2025, to 2.6 million now. Over the period, we have had to increase prices because of memory increases.

Speaker #4: That indeed, that demand was strongly left us with a backlog of units at the end of at the end of June 2026, compared to 600,000 at the end of 2025, to 2.6 million now.

Speaker #4: And similarly, over the period, we've had to increase prices because of memory increases. But pleased to see how the demand for products has held up through that period—and indeed, the demand for our 4-gigabyte and 8-gigabyte versions of Pi 5 and Pi 4 has continued to be very strong even though prices have had to go up.

Richard Boult: Pleased to see how the demand for products has held up through that period, and indeed, the demand for our 4 GB and 8 GB versions of Pi 5, Pi 4 has continued to be very strong, even though prices have had to go up. Overall, that gross profit per board improved significantly because of those price increases and also because of the benefit of memory that we had acquired at the end of 2025 at lower costs that we have used in that H1. Overall, we ended up with gross profit up to GBP 59 million, up from GBP 33 million a year ago, and that dropped through to a profit of adjusted EBITDA, our key profit metric, of GBP 40.3 million compared to GBP 19.4 million in the H1 of 2025. Does that work? Sorry, forgive me. It is a slight remote control whistle.

Speaker #4: Overall, that gross profit per board improved significantly because of those price increases, and also because of the benefit of memory that we had acquired at the end of 2025 at lower costs, which we used in that first half.

Speaker #4: So, overall, we ended up with gross profit up to £59 million, up from £33 million a year ago. And that dropped through to a profit of adjusted EBITDA, our key profit metric, of £40.3 million compared to £19.4 million in the first half of 2025.

Richard Boult: Overall, we ended up with gross profit up to GBP 59 million, up from GBP 33 million a year ago, and that dropped through to a profit of adjusted EBITDA, our key profit metric, of GBP 40.3 million compared to GBP 19.4 million in the H1 of 2025. Does that work? Sorry, forgive me. It is a slight remote control whistle. Moving on to the next slide. Yes. 26% growth in our direct units. Overall, the units supplied by us, so manufactured in our factory in Pencoed in Wales, grew by some 26%.

Speaker #4: Sorry, if you could give me a slight remote control with some... Moving on to the next slide. Yes, we had a 26% growth in our direct units.

Richard Boult: Moving on to the next slide. Yes. 26% growth in our direct units. Overall, the units supplied by us, so manufactured in our factory in Pencoed in Wales, grew by some 26%. The volumes with Farnell were a bit slower, and overall, we ended up with volumes up 17%. That was, again, as said earlier, particularly strong growth in Compute Modules and Pi 5, Pi 4. We saw lower growth, probably slightly negative actually in the case of Pico and Zero, which you may know are our lower cost boards. In the case of Zero, that volume was limited by access to supply. Our packager of the processor chip in Taiwan was somewhat distracted by data centers. That is now resolved, and that is part of the backlog reduction that we expect to see in the H2. Overall, strong performance on units.

Speaker #4: So overall, the units supplied by us—manufactured in our factory and with Sony in Wales—grew by some 26%. The volumes with Farnell were a bit slower.

Richard Boult: The volumes with Farnell were a bit slower, and overall, we ended up with volumes up 17%. That was, again, as said earlier, particularly strong growth in Compute Modules and Pi 5, Pi 4. We saw lower growth, probably slightly negative actually in the case of Pico and Zero, which you may know are our lower cost boards. In the case of Zero, that volume was limited by access to supply. Our packager of the processor chip in Taiwan was somewhat distracted by data centers. That is now resolved, and that is part of the backlog reduction that we expect to see in the H2. Overall, strong performance on units.

Speaker #4: And overall, we ended up with volumes up 17%. That was again said earlier, particularly strong growth in compute modules and Pi 5, Pi 4.

Speaker #4: We saw lower growth, probably slightly negative actually in the case of Pico in Xerox, you may know our lower cost boards. In the case of Xero, that volume was limited by access to supply.

Speaker #4: Our packager of the processor chip in Taiwan was somewhat distracted by data centers. That is now resolved, and that's part of the backlog reduction that we expect to see in the second half.

Speaker #4: So overall, strong performance on units. That translated into strong growth in revenue. With the increase in units, and also an increase in the average selling price—as I said, we increased our prices because we needed to pass through the increased costs of memory.

Richard Boult: That translated into a strong growth in revenue. With the increase in units and also an increase in the average selling price, as I said, we increased our prices because of needing to pass through the increased costs of memory. Our prices went up about 42%, leading to an over 90% increase in revenue through the period. Our revenue comprises a mixture of things. It includes royalty income for Farnell, it includes the direct sale to boards, it includes the sale of accessories, it also includes the sale of memory chips and processor chips to Farnell to make Raspberry Pis. It really does have a number of different things with quite different margin structures. When we look at the business, we focus more on gross profit and in particular, gross profit per unit combined with the number of units that we have sold.

Speaker #4: So, our prices went up about 42%, leading to an over 90% increase in revenue through the period. Our revenue comprises a mixture of things.

Speaker #4: It includes royalty income for Farnell. It includes the direct sales of boards. It includes the sale of accessories. It also includes the sale of memory chips and processor chips to Farnell to make Raspberry Pis.

Richard Boult: It includes royalty income for Farnell, it includes the direct sale to boards, it includes the sale of accessories, it also includes the sale of memory chips and processor chips to Farnell to make Raspberry Pis. It really does have a number of different things with quite different margin structures. When we look at the business, we focus more on gross profit and in particular, gross profit per unit combined with the number of units that we have sold.

Speaker #4: So it really does have a number of different things with quite different margin structures. So when we look at the business, we focus more on gross profit and, in particular, gross profit per unit combined with the number of units that we've sold.

Speaker #4: Gross profit overall increased by 79% due to those higher sales I’ve talked about, and also this improvement in gross profit per board. Gross profit per board—so that's the profit we're making from the sale of boards divided by the total number of boards in the period—increased by 53%, from $8 in the first half of last year to $12.20 in this half.

Richard Boult: Gross profit overall increased by 79% due to those higher sales I have talked about, and also this improvement in gross profit per board. Gross profit per board, so that is the profit we are making from the sale of boards divided by the total number of boards in the period, increased by 53% from $8 in H1 of last year to $12.20 in this H1. That reflected a combination of factors. It was the increase in the prices that we were selling and actually the way in which those took effect once we had announced them, but also this benefit of inventory that we brought into the beginning of the year where we had bought at significantly lower prices before the increase in memory costs that we have seen really in the last 6 to 9 months.

Speaker #4: That reflected a combination of factors. It was the increase in the prices that we were selling, and actually the way in which those took effect once we'd announced them, but also this benefit of inventory that we brought in at the beginning of the year where we had bought at significantly lower prices before the increase in memory costs that we've seen really in the last 6 to 9 months.

Speaker #4: That beneficial effect has probably been largely consumed now by the time of the middle of the year. I think it's fair to say that the current cost of inventory that we are now holding, we're now putting into boards, is much closer to the price at which we would buy further inventory at the moment.

Richard Boult: That beneficial effect has probably been largely consumed now by the time in the middle of the year. I think it is fair to say that the current cost of inventory that we are now holding, we are now putting into boards, is much closer to the price at which we would buy further inventory at the moment. The other standout was accessories. We had a very good H1 for accessories compared to the number of boards sold. It grew that gross profit per boards increased by, or sorry, the gross profit increased by 90% to $7.8 million compared to about half of that number a year ago. At a number of about $1.90 per board sold, which is well above our target that we set ourselves of $1 per board.

Speaker #4: The other standout was accessories. We had a very good half for accessories compared to the number of boards sold. It grew that gross profit per boards increased by well, sorry, the gross profit increased by 90% to 7.8 million dollars compared to about half of that number a year ago.

Speaker #4: That's a number of about $1.90 per board sold, which is well above our target that we set ourselves of a dollar per board. We saw particularly strong growth in compute in cameras.

Richard Boult: We saw particularly strong growth in cameras, in SD cards, displays, but also in AI HAT+ that we put together with our friends at Hailo, and we sell as an add-on to the Raspberry Pi. The Raspberry Pi is giving you additional compute particularly for AI models such as convolutional neural networks, or also for large language models on the latest version with Hailo. Overall, a strong growth of gross profit in the period. Moving on to the next slide. Just in putting those together in terms of how did our EBITDA evolve through the period. We came into H1 of last year was 19.4 million. Unit growth has given us probably nearly $5 million of additional profit. The additional units. That gross profit per board, so moving from $8 to $12 for each of those boards, is probably worth nearly $18 million of additional profit.

Speaker #4: In SD cards, displays, but also in AI hats that we put together with our friends at Halo—and we sell as an add-on to the Raspberry Pi—the Raspberry Pi is giving you additional compute, particularly for AI models such as convolutional neural networks, or also for large language models on the latest version with Halo.

Speaker #4: Overall, we saw strong growth of gross profit in the period. Moving on to the next slide, just putting those together in terms of how our EBITDA evolved through the period.

Richard Boult: Moving on to the next slide. Just in putting those together in terms of how did our EBITDA evolve through the period. We came into H1 of last year was 19.4 million. Unit growth has given us probably nearly $5 million of additional profit. The additional units. That gross profit per board, so moving from $8 to $12 for each of those boards, is probably worth nearly $18 million of additional profit.

Speaker #4: We came into the first half of last year at $19.4 million. Unit growth has given us probably nearly $5 million of additional profit.

Speaker #4: So the additional units—that gross profit per board, moving from $8 to $12 for each of those boards, is probably worth nearly $18 million of additional profit.

Speaker #4: Some of that is clearly because of the memory that we brought in, and also some orders that we had placed in the tail end of ’25, particularly at good prices this year.

Richard Boult: Some of that is clearly because of the memory that we brought in and also some orders that we had placed in the tail end of 2025 at particularly good prices this year. We think the value of that was probably a one-off, if you will, of somewhere between $10 or $15 million. Then the appreciation of accessories and then in terms of overheads, the research and development expenditure that we have that is not capitalized, it is research and development, it is research, it is software development, which typically we do not capitalize. That increased by about 700,000. That is predominantly some more heads, but also inflation salary increases. Then adjusted administrative costs. This is more of the general central team, be that sales, be it lotus eaters like myself. That grew by about $4.6 million. That was a combination of, there was some one-off effects of foreign exchange.

Speaker #4: We think the value of that was probably a one-off, if you will, of somewhere between 10 or 15 million dollars. And then the appreciation accessories.

Speaker #4: And then, in terms of overheads, the research and development expenditure that we have that is not capitalized—it's research and development, it's research, it's software development—which typically we don't capitalize.

Speaker #4: That increased by about 700,000. That's predominantly some more heads, but also inflation of salary increases. And then adjusted administrative costs. This is more of the general central team, be that sales, be it Lotus Eaters like myself. That grew by about $4.6 million.

Richard Boult: Then adjusted administrative costs. This is more of the general central team, be that sales, be it lotus eaters like myself. That grew by about $4.6 million. That was a combination of, there was some one-off effects of foreign exchange. There were some additional bonuses, reflecting this much stronger performance. There is about GBP 1 million of extra accrual, GBP 1.5 million there. The remainder is really headcount increases and increases in salaries generally across the team. Allowing for those, we ended with the H1 with a result of GBP 40.3 million of EBITDA.

Speaker #4: That was a combination of there was some one-off effects of foreign exchange. There was some additional bonuses. Reflecting this much stronger performance. That was probably there's about a million of extra accrual, a million and a half there.

Richard Boult: There were some additional bonuses, reflecting this much stronger performance. There is about GBP 1 million of extra accrual, GBP 1.5 million there. The remainder is really headcount increases and increases in salaries generally across the team. Allowing for those, we ended with the H1 with a result of GBP 40.3 million of EBITDA. The next slide sets out the income statement. Very much the matters we have just covered recently. I am going to skip over that one. I am going to move on to the description of the balance sheet here, but I think the best way to express that is in the cash flow and how our cash flow has moved across the period. We came into the beginning of this year with $28 million of cash. EBITDA, the profits that we have just been talking about, contributed $40 million. We have then spent significantly on inventory.

Speaker #4: And the remainder is really headcount increases and increases in salaries generally across the team. Allowing for those, we ended the half with a result of £40.3 million.

Speaker #4: EBITDA. The next slide sets out the income statement. Very much the matters we've just covered recently. I'm going to skip over that one, and I am going to move on to the description of the balance sheet here, but I think the best way to express that is in the cash flow, and how our cash flow has moved across the period.

Richard Boult: The next slide sets out the income statement. Very much the matters we have just covered recently. I am going to skip over that one. I am going to move on to the description of the balance sheet here, but I think the best way to express that is in the cash flow and how our cash flow has moved across the period. We came into the beginning of this year with $28 million of cash. EBITDA, the profits that we have just been talking about, contributed $40 million. We have then spent significantly on inventory.

Speaker #4: We came into the beginning of this year with $28 million of cash. The EBITDA, the profits that we've just been talking about, contributed $40 million.

Speaker #4: We've then spent significantly on inventory. That increase in inventory that we are now carrying is almost entirely due to purchases of DRAM that we use in our boards—£106 million of that is DRAM.

Richard Boult: That is almost entirely that increase in inventory that we are now carrying is $106 million of that is purchases of DRAM that we use in our boards. That is the position at the end of June. That has put us into a much stronger position in terms of having nearly 3 months of inventory. We have continued to add to that recently such that we have good confidence that we can meet our production plans for the remainder of this year and into the Q1 of next year. That memory obviously these days is at a significantly higher price than it was a few years ago. That inventory at the end of June was funded in part by payables. We bought significant amounts in June itself, and therefore there was a payable at the end of the period. That has started to reduce.

Speaker #4: I think that has and that is the position at the end of June. That's put us into a much stronger position in terms of having nearly three months of inventory.

Speaker #4: We've continued to add to that recently, such that we have good confidence that we can meet our production plans for the remainder of this year.

Speaker #4: And into the first quarter of next year. But that memory, obviously these days, is at a significantly higher price than it was a few years ago.

Speaker #4: That inventory at the end of June was funded in part by payables. We bought significant amounts in June itself, and therefore, there was a payable at the end of the period.

Richard Boult: We bought significant amounts in June itself, and therefore there was a payable at the end of the period. That has started to reduce. This is where you have seen a cash outflow continuing post the end of June, and therefore we are drawing on our bank facilities. We increased those facilities to GBP 140 million of committed funds.

Speaker #4: That has started to reduce, so we've seen a cash outflow continuing post the end of June. Therefore, we're drawing on our bank facilities.

Richard Boult: This is where you have seen a cash outflow continuing post the end of June, and therefore we are drawing on our bank facilities. We increased those facilities to GBP 140 million of committed funds. That was completed in early July, which has given us a position where we are able to buy inventory where the opportunity arises to secure that future production, to secure good prices on the products such that we can maintain an attractive level of profitability. But I think it is fair to say that in the H2, the profit per unit will come down as those memory costs now are closer to market levels rather than based on prices from very good historic purchases. Receivables, also, we saw some outflow in those as at the end of June.

Speaker #4: We increased those facilities to $140 million. I've committed funds that were completed in early July, which has given us a position where we're able to buy inventory when the opportunity arises.

Richard Boult: That was completed in early July, which has given us a position where we are able to buy inventory where the opportunity arises to secure that future production, to secure good prices on the products such that we can maintain an attractive level of profitability. But I think it is fair to say that in the H2, the profit per unit will come down as those memory costs now are closer to market levels rather than based on prices from very good historic purchases. Receivables, also, we saw some outflow in those as at the end of June.

Speaker #4: To secure that future production to secure good prices on the products such that we can maintain an attractive level of profitability I think it's fair to say that in the second half, the profit per unit will come down as those memory costs now are closer to market levels, rather than based on prices from very good historic purchases.

Speaker #4: Receivables also—we saw some outflow in those as at the end of June. That was really a function of just how strong trading in the month of June was.

Richard Boult: That was really a function of just how strong trading in the month of June was, very active month, obviously compared to December, which is a quieter month by the end of the month itself. But also just generally, trading has been so much stronger with real momentum coming into this H2. We spent about GBP 11 million on CapEx in the H1, very consistent with the guidance that we have given for the last few years of about GBP 20 million of CapEx for any full year. So that continues. A little bit of tax, leaving us with a closing position of about GBP 8.4 million of tax. As I say, I think in the H2, those payables will reduce. They were particularly high at the end of June. We also expect to be buying and holding quite a bit of memory.

Speaker #4: It’s been a very active month, obviously, compared to December, which is a quieter month by the end of the month itself. But also, just generally, trading has been much stronger, with real momentum coming into this second half.

Speaker #4: We spent about $11 million on capex in the first half. Very consistent with the guidance that we've given for the last few years of about $20 million of capex for any full year, so that continues.

Speaker #4: A little bit of tax, leaving us with a closing position of about £8.4 million of tax. As I say, I think in the second half, those payables will reduce.

Speaker #4: They were particularly high at the end of June. And we also expect to be buying and holding quite a bit of memory as we can see the opportunity to do so.

Richard Boult: They were particularly high at the end of June. We also expect to be buying and holding quite a bit of memory. We can see the opportunity to do so. That will leave us probably with acquiring further inventory, such that we would expect to see our debt levels at, say, GBP 40 to GBP 50 million. I think we should flag that if opportunities arise, if there are good prices, if there is the opportunity to acquire the right memory, we will utilize some of the additional headroom we have on the bank facilities to add to that inventory to make sure that we go into next year in a particularly strong shape.

Richard Boult: We can see the opportunity to do so. That will leave us probably with acquiring further inventory, such that we would expect to see our debt levels at, say, GBP 40 to GBP 50 million. I think we should flag that if opportunities arise, if there are good prices, if there is the opportunity to acquire the right memory, we will utilize some of the additional headroom we have on the bank facilities to add to that inventory to make sure that we go into next year in a particularly strong shape. Just touching on a final slide from me. Memory purchases. I think one of the things that is notable is compared to the announcement that we made in early June about how strong trading was, a lot of the numbers I have talked about here are very much in line with that.

Speaker #4: That will leave us, probably, with acquiring further inventory, such that we would expect to see our debt levels at, say, $40 to $50 million.

Speaker #4: But I think we should flag that if opportunities arise, if there are good prices, if there is the opportunity to acquire the right memory, we will utilize some of the additional headroom we have on the bank facilities to add to that inventory.

Speaker #4: To make sure that we go into next year in a particularly strong shape. Just touching on a final slide from me—memory purchases. I think one of the things that's notable is, compared to the announcement that we made in early June about how strong trading was, a lot of the numbers I've talked about here are very much in line with that.

Richard Boult: Just touching on a final slide from me. Memory purchases. I think one of the things that is notable is compared to the announcement that we made in early June about how strong trading was, a lot of the numbers I have talked about here are very much in line with that. I think to me, the additional feature, the additional new news probably is the strength of our position with memory purchases. I think Mike Buffham, our Chief Commercial Officer, has done a phenomenal job of talking to packagers, manufacturers of memory to secure that. That position, I think, is a step forwards from where we were even two or three months ago.

Speaker #4: I think, to me, the additional feature, the additional new news, probably is the strength of our position with memory purchases. I think Mike Buffer, my Chief Commercial Officer, has done a phenomenal job of talking to package manufacturers of memory to secure that.

Richard Boult: I think to me, the additional feature, the additional new news probably is the strength of our position with memory purchases. I think Mike Buffham, our Chief Commercial Officer, has done a phenomenal job of talking to packagers, manufacturers of memory to secure that. That position, I think, is a step forwards from where we were even two or three months ago. How have we got into that position with a stronger position on memory? We have diversified suppliers. We now have approximately seven different companies providing memory to us compared to two a year ago. We have used our engineering skills to technically innovate, to actually allow us to use two 1 gigabyte pieces of DRAM to create a 2 gigabyte board. If I make sure I got the maths right on that. Two lots of two to make four, which, gosh, it is all binary. How easy.

Speaker #4: And that position, I think, is a step forwards from where we were even two or three months ago. How have we got into that position with a stronger position on memory?

Richard Boult: How have we got into that position with a stronger position on memory? We have diversified suppliers. We now have approximately seven different companies providing memory to us compared to two a year ago. We have used our engineering skills to technically innovate, to actually allow us to use two 1 gigabyte pieces of DRAM to create a 2 gigabyte board. If I make sure I got the maths right on that. Two lots of two to make four, which, gosh, it is all binary. How easy.

Speaker #4: We've diversified suppliers. We now have approximately seven different companies providing memory to us compared to two a year ago. We have used our engineering skills to technically innovate, to actually allow us to use two one gigabyte pieces of DRAM to create a two gigabyte board for make sure I got the maths right on that.

Speaker #4: And two lots of two to make four, which, gosh, it's all binary. How easy. Similarly, on pricing, we have I think minimized the disruption to our customers' expectations of us by communicating quite clearly the reasons for price increases to be seen to be passing through price increases rather than gouging.

Richard Boult: Similarly on pricing, we have, I think, minimized the disruption to our customers' expectations of us by communicating quite clearly the reasons for price increases to be seen to be passing through price increases rather than gouging on selected products and really also providing in terms of a Raspberry Pi 3 gigabyte variant, additional options for customers who maybe did not want to go to suffer the whole of the price increase that was necessary. We have continued to make strategic purchases. We have seen the opportunity with our balance sheet, with our connections, our reputation to make further purchases from suppliers who in certain cases are keen to establish a presence elsewhere. We have been seen to be a very good partner to them in achieving that.

Speaker #4: I've selected products and, really, also providing, in terms of a Raspberry Pi three-gigabyte variant, additional options for customers who maybe did not want to go to suffer the whole of a price increase that was necessary.

Speaker #4: And we've continued to make strategic purchases. We have seen the opportunity with our balance sheet, with our connections, our reputation, to make further purchases from suppliers who in certain cases are keen to establish a presence elsewhere and we have been a very good we have been seen to be a very good partner to them in achieving that.

Richard Boult: We have seen the opportunity with our balance sheet, with our connections, our reputation to make further purchases from suppliers who in certain cases are keen to establish a presence elsewhere. We have been seen to be a very good partner to them in achieving that. I think overall, a lot of activity in terms of securing memory, and that has put us into a stronger position than we were, say, three months ago. I think in that good position, hand over to Evan to talk some more.

Speaker #4: So I think overall, a lot of activity in terms of securing memory. And that's put us into a stronger position than we were, say, three months ago.

Richard Boult: I think overall, a lot of activity in terms of securing memory, and that has put us into a stronger position than we were, say, three months ago. I think in that good position, hand over to Evan to talk some more.

Speaker #4: But I think we're in that good position, and I'll hand it over to Evan to talk some more.

Speaker #1: Thank you, Richard. So before we dive into the before we dive into the sort of the strategy update, just a recap about the three pillars of our growth strategy.

Eben Upton: Thank you, Richard. Before we dive into the strategy update, just to recap about the three pillars of our growth strategy. Firstly, obviously to grow unit sales through investment in channel, and increasingly through our board-to-board program through direct outreach to OEMs. Then for each of those units we sell, aiming to grow the unit profit margin. That is both through engineering to grind cost out of the platform, and over time to acquire a larger share of the silicon build of materials in each device. The gradual migration, we have seen this continue this year, the gradual migration of our customers to higher density SKUs. Even in this constrained DRAM environment, we have seen an increasing preference from many of our customers for higher density SKUs. Those are higher density SKUs, they are higher ASP SKUs, and that higher ASP brings with it a higher unit profit.

Speaker #1: Firstly, obviously, to grow unit sales, through investment in channel. And through increasingly through our board-to-board program, through direct outreach to OEMs. Then for each of those units we sell, aiming to grow the unit profit margin, that's both through engineering to grind cost out of the platform.

Speaker #1: And, over time, to acquire a larger share of the silicon bill of materials in each device. And the gradual migration—and we've seen this continue this year—the gradual migration of our customers to higher-density SKUs, even in this constrained DRAM environment.

Eben Upton: The gradual migration, we have seen this continue this year, the gradual migration of our customers to higher density SKUs. Even in this constrained DRAM environment, we have seen an increasing preference from many of our customers for higher density SKUs. Those are higher density SKUs, they are higher ASP SKUs, and that higher ASP brings with it a higher unit profit.

Speaker #1: We have seen an increasing preference from many of our customers for higher-density SKUs. Those are higher-density SKUs; they're higher ASP SKUs, and that higher ASP brings with it a higher unit profit.

Speaker #1: And then finally, growing our margin participation, growing our participation in that margin. Obviously, the board-to-board initiative particularly fertile source of direct relationships with OEMs of our three different routes to market.

Eben Upton: Finally, growing our margin participation, growing our participation in that margin. Obviously, the board-to-board initiative, particularly fertile source of direct relationships with OEMs of our three different routes to market. The direct to OEM route is the route that allows us to capture the largest share of that unit profit margin in the device. A little bit more detail about product launches. As we said, five launches in the half. That is a relatively quiet half for us compared to recent years. But with some real strategic highlights that I think speak to the strategic direction of the business. At the start of the half, the second generation Raspberry Pi AI Kit accessory, co-developed with our friends at Hailo, and bringing generative AI capabilities to the platform.

Speaker #1: The direct-to-OEM route is the route that allows us to capture the largest share of that unit profit margin in the device. A little bit more detail about product launches.

Speaker #1: As we said, five launches in the half. That is a relatively quiet half for us compared to recent years. But with some real strategic highlights that I think speak to the strategic direction of the business.

Speaker #1: At the start of the half, the second-generation Raspberry Pi AI Hat accessory, co-developed with our friends at Halo, is bringing generative AI capabilities to the platform.

Speaker #1: The original AI accessory was very much focused on vision operations and classical network architectures, specifically convolutional neural network architectures. This new product brings with it both the ability to run small or large language models, as well as vision-language models.

Eben Upton: The original AI accessory very much focused on vision operations, very much focused on the classical network architectures, so convolutional neural network architectures. This new product bringing with it both the ability to run small or large language models and vision language models, and also on the vision side then for modern architectures, so transformer-based, vision transformer-based classification algorithms. In the middle of the period, the 3 GB variant of Raspberry Pi 4, that is in direct response to specific OEM requests for an intermediate density point. We continue to be surprised by the extent to which our OEM customers are not trading down. If anything, they are trading up for memory.

Speaker #1: And also on the vision side, then more modern architecture. So transformer-based vision, transformer-based classification algorithms. In the middle of the period, the three gigabyte variant of Raspberry Pi 4.

Speaker #1: That's in direct response to specific OEM requests for an intermediate density point. We continue to be surprised by the extent to which our OEM customers aren't trading down.

Speaker #1: If anything, they're trading up for memory. But we do have OEM customers who welcome the opportunity to invest fixed costs, to do engineering, to make engineering investments to take memory requirements out of their software platforms.

Eben Upton: But we do have OEM customers who welcome the opportunity to invest fixed costs to make engineering investments to take memory requirements out of their software platforms, providing for, if you think the 4 GB products, really they are the heart in our modern products, the Raspberry Pi 4 and Raspberry Pi 5, 4 GB really is the heart of the business. We have an enormous number of 4 GB OEM customers. It is challenging for a 4 GB customer to come down to 2 GB by providing that intermediate density point on the Raspberry Pi 4 platform. We are able to serve those customers who want to trade engineering effort to unit cost. They are not quite in the half, just off the end, just into the start of July at the start of this half. We have the programming jig for Compute Module 5.

Speaker #1: Providing for, if you think of four gig products, really they're at the heart of our modern products—the Raspberry Pi 4 and Raspberry Pi 5.

Speaker #1: Four gigabytes, really, is the heart of the business. We have an enormous number of four gigabytes OEM customers. It is challenging for a four gigabyte customer to come down to two gigabytes by providing that intermediate density point on the Raspberry Pi 4 platform.

Eben Upton: We have an enormous number of 4 GB OEM customers. It is challenging for a 4 GB customer to come down to 2 GB by providing that intermediate density point on the Raspberry Pi 4 platform. We are able to serve those customers who want to trade engineering effort to unit cost. They are not quite in the half, just off the end, just into the start of July at the start of this half. We have the programming jig for Compute Module 5.

Speaker #1: We're able to serve those customers who want to trade engineering effort for unit cost. And then, not quite in the half—just off the end, just into the start in July, at the start of this half—we have the programming jig for Compute Module 5.

Speaker #1: This is our part of our offering to our OEM customers who wish to be able to program Raspberry Pi compute modules in production at scale.

Eben Upton: This is part of our offering to our OEM customers who wish to be able to program Raspberry Pi Compute Modules in production at scale, designed to integrate with some of the work we have done on our operating system assembly and imaging over the last couple of years. In parallel with all these efforts, of course, we have seen a series of upgrades to Raspberry Pi Connect, our IT cloud platform. Very different cadence for this from what I am certainly used to. Certainly, if you imagine the cadence of hardware releases being on the order of years, the cadence of operating system software releases being on the order of quarters or months. This is a product which has seen multiple upgrades with a cadence of weeks across the H1.

Speaker #1: Designed to integrate with some of the work we've done, and our operating system assembly and imaging, over the last couple of years. In parallel with all these efforts, of course, we've seen a series of upgrades to Raspberry Pi Connect, our IoT cloud platform.

Speaker #1: Very different cadence for this from what I'm certainly used to. Certainly, if you imagine the cadence of hardware releases being on the order of years, the cadence of operating even of operating system software releases being on the order of months.

Speaker #1: Quarters of months. This is a product which has seen multiple upgrades with a cadence of weeks across the half. Seeing continued that's driving continued strong growth in adoption, both of the base one Raspberry Pi Connect platform, the free Raspberry Pi Connect platform, and the paid for Raspberry Pi Connect for organizations platform.

Eben Upton: That is driving continued strong growth in adoption, both of the baseline Raspberry Pi Connect platform, the free Raspberry Pi Connect platform, and the paid for Raspberry Pi Connect for Organisations platform. We love case study slides. We could not not feature this one, I think in this slide deck. Formlabs, a 3D printing company. They are a long time Raspberry Pi adopter, very loyal adopter of Raspberry Pi technology, outsourcing, as with many of our OEM customers, outsourcing really the intelligence element, the compute element at the heart of their professional 3D printers to us. A wonderful gesture. Every time an OEM makes that choice, it is a gesture of commitment. It is a gesture of faith in the Raspberry Pi organization and the Raspberry Pi platform.

Speaker #1: Now, we love case study slides. We couldn't not feature this one, I think, in the slide deck. Formlabs 3D printer company. They're a longtime Raspberry Pi adopter, a very loyal adopter of Raspberry Pi technology.

Speaker #1: Outsourcing, as with many of our OEM customers, really concerns the intelligence element—the compute element—at the heart of their professional 3D printers. To us, it's a wonderful gesture, I guess; every time an OEM makes that choice, it is a gesture of commitment.

Speaker #1: It's a gesture of faith in the Raspberry Pi organization and the Raspberry Pi platform. And it was wonderful to see the Formlabs Raspberry Pi–powered Formlabs 3D printer taking pride of place in one of the videos that was shown at this month's Apple event.

Eben Upton: It was wonderful to see a Raspberry Pi powered Formlabs 3D printer taking pride of place in one of the videos that was shown at this month's Apple event. So these devices really do get absolutely everywhere. Really, this has been a year, we talked a little bit about products, but really, I think the news about this H1 for us is it has been a period of organizational transformation for Raspberry Pi. We have been growing, we have been evolving the executive team inside the organization. in March, we appointed Tim Mamtora, a very old friend of mine, as our very first Chief Operating Officer. This is Richard's last appearance in a results presentation. Tim Powell will be succeeding Richard as Chief Financial Officer from the end of next month.

Speaker #1: So these devices really do get absolutely everywhere. Really, this has been a year we talked a little bit about products. But really, I think the news about this half for us is it's been a period of organizational transformation for Raspberry Pi.

Speaker #1: We have been growing. We've been evolving. The executive team inside the organization in March, we appointed Tim Mantora, a very old friend of mine, as our very first chief operating officer.

Speaker #1: And then this is Richard's last appearance in a results presentation. Tim Powell will be succeeding Richard as chief financial officer from the end of next month.

Speaker #1: We've been making targeted hires across the rest of the organization as well. In engineering, in operations, in finance, and in enterprise sales as we aim to both increase the organization's ability to develop innovative new products and then our ability to support those to sell and support those products into our enthusiast our industrial and our OEM customer base.

Eben Upton: We have been making targeted hires across the rest of the organization as well in engineering, in operations, in finance, and in enterprise sales as we enter both increase the organization's ability to develop innovative new products and then our ability to sell and support those products into our enthusiast, our industrial, and our OEM customer base. In the new operations organization under Tim, there has been a focus really on building the capabilities, the structures and capabilities that we need to support the next phase of Raspberry Pi's growth.

Speaker #1: There's been a focus in the new operations organization under Tim. There's been a focus really on building the capabilities, the structures and capabilities that we need to support the next phase of Raspberry Pi's growth.

Speaker #1: This is across the organization in engineering and operations, but with a particular focus on manufacturing and supply chain operations. We're really driving solely with our manufacturing partner for the majority of our products.

Eben Upton: This is across the organization in engineering operations, but with a particular focus on manufacturing and supply chain operations, really driving Sony, our manufacturing partner, for the majority of our products, driving their production capacity out to 1 million units a month, which is where we believe it needs to be in order to support our growth ambitions over the next few years. We have been expanding the commercial team as well, both colleagues who support our approved reseller and authorized distributor channel, and the partners who support the growing number of direct OEMs. As we have deepened our direct relationship with OEMs, particularly with the OEMs who we meet through our board-to-board program, these tend to be larger potential opportunities, but they also tend to be opportunities which may require more targeted support from our organization to get those opportunities over the line.

Speaker #1: Driving their production capacity out to one million units a month, which is where we believe it needs to be in order to support our growth ambitions over the next few years.

Speaker #1: We've been expanding the commercial team as well, including colleagues who support our approved reseller and authorized distributor channels, and partners who support the growing number of direct OEM engagements within the business.

Eben Upton: We have been expanding the commercial team as well, both colleagues who support our approved reseller and authorized distributor channel, and the partners who support the growing number of direct OEMs. As we have deepened our direct relationship with OEMs, particularly with the OEMs who we meet through our board-to-board program, these tend to be larger potential opportunities, but they also tend to be opportunities which may require more targeted support from our organization to get those opportunities over the line.

Speaker #1: As we've deepened our direct relationship with OEMs, particularly with the OEMs whom we meet through our board-to-board program, these tend to be larger potential opportunities.

Speaker #1: But they also tend to be opportunities which may require more targeted support from our organization to get those opportunities over the line. So we've been making those investments in our application engineering team to sustain to be able to sustain those engagements even as they require a little bit more touch as I say to get them over the line.

Eben Upton: We have been making those investments in our application engineering team to be able to sustain those engagements, even as they require a little bit more touch, as I say, to get them over the line. A word on competitive landscape. It really is a rapidly changing competitive environment at the moment. I think what we are seeing is we are seeing Raspberry Pi's core brand values, our very long-standing core brand values, our industry-leading cost structure, which is underpinned both by proprietary technology and by that end-to-end engineering culture we have built, really can go hunting for cost optimization opportunities throughout the entire technology stack. Our commitments to the very long-term support and very long-term availability of our products, and the unmatched organic ecosystem that has grown up around Raspberry Pi.

Speaker #1: A word on the competitive landscape. It really is a rapidly changing competitive environment at the moment. And I think what we're seeing is Raspberry Pi's core brand values—our very longstanding core brand values—our industry-leading cost structure, which is underpinned both by proprietary technology and by that end-to-end engineering culture we've built, really can go hunting for cost optimization opportunities throughout the entire technology stack.

Speaker #1: Our commitments to the very long-term support and very long-term availability of our products, and the unmatched organic ecosystem that's grown up around Raspberry Pi.

Speaker #1: So these are our longstanding brand values. And I think what we're seeing this year, and what we've seen over the last couple of years, is these starting to intersect with a set of new imperatives in this more complex environment we're living in.

Eben Upton: These are our long-standing brand values, and I think what we are seeing this year, and we have seen over the last couple of years, is these starting to intersect with a set of new imperatives in this more complex environment we are living in. We are seeing a greater emphasis in many of our markets on sovereign capability. For our American customers, North America, United States, accounts for roughly a third of our business. In that environment, our customers have a particular focus now on the tariff implications of the tariff behavior of their supply chain. There we are building our products, building the vast majority of our products in the United Kingdom. We are differentially advantaged from a tariff perspective versus almost any other non-domestic manufacturing location for an American customer.

Speaker #1: So we're seeing a greater emphasis in many of our markets on sovereign capability. For our American customers, North America, the United States, accounts for roughly a third of our business.

Speaker #1: And in that environment, our customers have a particular focus now on the tariff implications of the tariff behavior of their supply chain. And there we are building our products, building the fast majority of our products in the United Kingdom.

Speaker #1: We are differentially advantaged from a tariff perspective versus almost any other non-domestic manufacturing location for an American customer. And finally, we see a growing amount of regulatory burden—a growing regulatory burden—on our customers who are building connected products.

Eben Upton: Finally, we see a growing amount of regulatory burden, a growing regulatory burden on our customers who are building connected products. Our customers who are building IoT products are having to comply with an increasingly strict set of requirements in order to be able to legally ship those, duly the Cyber Resilience Act as the most obvious example. It is increasingly important not just to be able to build products, but to keep those products secure over time, to have a story about how you keep them secure over time, an evidenced story about how you expect to keep those products secure over time. You are not just generating when you build a software image for your system, you are not just generating software, of course, you are also generating compliance artifacts which must be synchronized with that software build.

Speaker #1: Our customers who are building IoT products, are having to comply with an increasingly strict set of requirements in order to be able to legally ship those.

Speaker #1: The Cyber Resilience Act is the most obvious example. It is increasingly important not just to be able to build products, but to keep those products secure over time—to have a story about how you keep them secure over time, an evidence story about how you expect to keep those products secure over time.

Speaker #1: And you're not just generating when you build a software image for your system. You're not just generating software. Of course, you are also generating compliance artifacts, which must be synchronized with that software build.

Speaker #1: And many of the investments that we've made over the last few years in deepening the Raspberry Pi software ecosystem are really coming to bear and are becoming increasingly salient as people as many of our OEM customers, many of our IoT focused OEM customers are confronting some of those challenges.

Eben Upton: Many of the investments that we have made over the last few years in deepening the Raspberry Pi software ecosystem are really coming to bear and are becoming increasingly salient as people, as many of our OEM customers, many of our IoT focused OEM customers are confronting some of those challenges. All of this is happening in a much more challenging supply chain environment. I think we are seeing signs this year, one of the drivers of growth in sales, one of the drivers of growth in that backlog, has been what we might call a flight to quality, that we have OEM customers we are meeting just as our existing OEM customers continue to scale, and as the pipeline that we have brought with us into the year continues to mature.

Speaker #1: And of course, all of this is happening in a much more challenging supply chain environment. I think we are seeing signs this year one of the drivers of growth in sales one of the drivers of growth in that backlog has been what we might call a flight to quality.

Speaker #1: That we have OEM customers we are meeting just as our existing OEM customers continue to scale and as the pipeline that we've brought with us into the year continues to mature.

Speaker #1: We are also seeing new OEM engagements which are driven by the by our OEMs who have designed with other modular platforms. OEMs who have chosen to outsource compute to other vendors, where those vendors are now struggling to keep their products in production.

Eben Upton: We are also seeing new OEM engagements which are driven by our OEMs who have designed with other modular platforms, OEMs who have chosen to outsource compute to other vendors where those vendors are now struggling to keep their products in production, or OEMs who have chosen the in-house engineering solution, OEMs who have chosen to make rather than to buy, those in-house teams may also be struggling to secure the components they need. We are seeing people coming to us as a trusted partner to see them through this supply chain environment. For us, that is an opportunity to meet new customers. It is an opportunity to gain the faith of new customers, to be of service to those customers, and hopefully many of those customers will remain with us even after the present supply chain environment has improved a little bit.

Speaker #1: Or OEMs who have chosen the in-house engineering solution—OEMs who have chosen to make rather than to buy. Those in-house teams may also be struggling to secure the components they need.

Speaker #1: So, we are seeing people come to us as a trusted partner to see them through this supply chain environment. And, of course, for us, that is an opportunity to meet new customers.

Eben Upton: We are seeing people coming to us as a trusted partner to see them through this supply chain environment. For us, that is an opportunity to meet new customers. It is an opportunity to gain the faith of new customers, to be of service to those customers, and hopefully many of those customers will remain with us even after the present supply chain environment has improved a little bit.

Speaker #1: It's an opportunity to get in the face of new customers, to be of service to those customers, and hopefully many of those customers will remain with us even after the present supply chain environment has improved a little bit.

Speaker #1: A few brief words on AI. We all know that we are living in an age of AI-related disruption. This impacts our business in numerous ways.

Eben Upton: A few brief words on AI. We all know that we are living in an age of AI-related disruption. This couples into our business in numerous ways. Obviously, we have already talked about some of the supply chain impacts of competing with the hyperscaler AI build-out for access to a number of components, most notably storage components, volatile and non-volatile memory. I think the more we see so many of our customers, the applications that our IoT OEM customers are running have a very strong AI flavor to them. The more we see of how our customers are using AI, the more convinced we become that there is a significant opportunity for us here. We believe the thesis is that there is a centrifugal tendency over time, a tendency for AI compute to migrate from the center of the network.

Speaker #1: Obviously, we've already talked about some of the supply chain impacts of competing with the AI, the hyperscaler AI build-out for access to a number of components, most notably storage components, volatile and non-volatile memory.

Speaker #1: But I think the more we see, so many of our customers—the applications of our IoT OEM customers—are running and have a very strong AI flavor to them.

Speaker #1: And the more we see of how our customers are using AI, the more convinced we become that there is a significant opportunity for us here.

Speaker #1: We believe the thesis is that there is a centrifugal tendency over time—a tendency for AI compute to migrate from the center of the network. Any given AI technology, any given AI technique, may start at the center of the network, and over time, as the devices at the edge of the network become more performant, this changes.

Eben Upton: Any given AI technology, any given AI technique may start at the center of the network, and over time, as the devices at the edge of the network become more performant and as research effort in AI drives down the compute requirement of a specific level of performance, that computer that starts from the center of the network will naturally migrate to the edge. As it migrates to the edge, that migration brings with it improvements in reliability, in cost structure, in privacy, and in security. This does feel like a general rule. We absolutely see it today in visual AI.

Speaker #1: And as research effort in AI drives down the compute requirement of a specific level of performance, that compute that starts at the center of the network will naturally migrate to the edge, bringing and as it migrates to the edge, that migration brings with it improvements in reliability, in cost structure, in privacy, and in security.

Speaker #1: This does feel like a general rule. We absolutely see it today in visual AI. Many of our largest, many of our highest-volume OEM customers today are using Raspberry Pi platforms to run visual AI algorithms.

Eben Upton: Many of our largest, many of our highest volume OEM customers today are using Raspberry Pi platforms to run visual AI algorithms, which would have been inconceivable to run at the edge of the network a decade ago. We are seeing the same thing already happen in generative AI. We are seeing a collapse in the compute requirement of a given level of generative AI performance. We believe that transition from the center to the edge of the network, which has taken roughly 10 years in previous AI applications, in visual AI applications, is likely to take place over the next 2 or 3 or 4 years in generative. By the end of the decade, the vast majority of inference operations for generative AI will happen at the edge rather than at the core of the network.

Speaker #1: Which would have been inconceivable to run at the edge of the network a decade ago. We are seeing the same thing already happen in generative AI.

Speaker #1: We are seeing a collapse in the compute requirement of a given level of generative AI performance. And so we believe that that transition from the center to the edge of the network, which is taking roughly 10 years in the previous AI applications, in visual AI applications, is likely to take place over the next two or three or four years.

Speaker #1: In generative, by the end of the decade, the vast majority of inference operations for generative AI will happen at the edge rather than at the core of the network.

Speaker #1: These AI these trends in AI are driving demand across a very broad range of markets. I'm going to call out two today. Those being smart home and aerospace and events.

Eben Upton: These trends in AI are driving demand across a very broad range of markets. I am going to call out 2 today. Those being smart home and aerospace and defense. Some words on smart homes, smart environments. Customers like Homey are using Raspberry Pi technology, another very longstanding, extremely loyal Raspberry Pi customer. They are using the Raspberry Pi platform to build intelligence into our everyday interactions with our environment. I love this quote from Emil, "We didn't have to reinvent the wheel." When he says that, really what he is doing there is he is articulating the core of the Raspberry Pi value proposition to our OEM customers, that you do not have to become a computer company if you want to embed intelligence into a product at the edge of the network. You do not have to, you should not have to become a computer company overnight.

Speaker #1: So, some words on smart home and smart environments. Customers like Homey are using Raspberry Pi technology—another very longstanding, extremely loyal Raspberry Pi customer. They're using the Raspberry Pi platform to build intelligence into our everyday interactions with our environments.

Eben Upton: They are using the Raspberry Pi platform to build intelligence into our everyday interactions with our environment. I love this quote from Emil, "We didn't have to reinvent the wheel." When he says that, really what he is doing there is he is articulating the core of the Raspberry Pi value proposition to our OEM customers, that you do not have to become a computer company if you want to embed intelligence into a product at the edge of the network. You do not have to, you should not have to become a computer company overnight.

Speaker #1: I love this quote from Emil: "We didn't have to reinvent the wheel." And he says that really what he's doing there is articulating the core of the Raspberry Pi value proposition to our OEM customers.

Speaker #1: That you don't have to become a computer company. If you want to embed intelligence into a product at the edge of the network, you shouldn't.

Speaker #1: You don't have to. You should not have to become a computer company. You should be able to find something at your house whilst that too.

Eben Upton: You should be able to find somebody to outsource that to. It is our aspiration to win that argument. When we have won that argument, then to win the argument that the natural person to outsource this to is Raspberry Pi. This outsourcing is, I think, a very long-term trend that has been to our benefit over the last few years, and it has been accelerated by the challenges around memory pricing and availability, as I have said, by some of the challenges around regulatory complexity, and of course, lastly, by the challenges associated with acquiring embedded engineering talent. Many of our OEM customers are coming to us for compute solutions in order to avoid not just the cost, but the administrative complexity of building and maintaining a team which can build and maintain in production the compute elements at the heart of their platform.

Speaker #1: It's our aspiration to win that argument. And when we've won that argument, then to win the argument that the natural person to ourselves is Raspberry Pi.

Speaker #1: This outsourcing is, I think, a very long-term trend that's been to our benefit over the last few years. And it's been accelerated by the challenges around memory pricing and availability, as I've said, and by some of the challenges around regulatory complexity.

Speaker #1: And of course, lastly, by the challenges associated with acquiring embedded engineering talent. Many of our OEM customers are coming to us for compute solutions in order to avoid the not just the cost, but the administrative complexity of building and maintaining a team, which can build and maintain in production the compute elements at the heart of their platform.

Eben Upton: Many of our OEM customers are coming to us for compute solutions in order to avoid not just the cost, but the administrative complexity of building and maintaining a team which can build and maintain in production the compute elements at the heart of their platform. We are allowing people to focus on their differentiating, on their value add, on their differentiating engineering, not on the non-differentiating engineering of being a computer company. Defense, becoming an increasingly important potential market for us.

Speaker #1: We're allowing people to focus on their value-add, on their differentiating engineering, not on the non-differentiating engineering of being a computer company.

Eben Upton: We are allowing people to focus on their differentiating, on their value add, on their differentiating engineering, not on the non-differentiating engineering of being a computer company. Defense, becoming an increasingly important potential market for us. There are a number of secular demand drivers going on, obviously here in the background, a trend in the West towards rearmament, a recognition that we have probably wound down the size of our armed forces further than is wise. A growing recognition, driven probably by some of the experiences that are happening today in Ukraine, of the importance of autonomy, of advanced autonomy, of distributed decision-making, in mobile platforms. A pivot away, I think, from what we might call exquisite platforms, towards high cost, low volume platforms with very long development times, with development times on the order of decades, towards lower cost, attritable, lower cost consumable platforms.

Speaker #1: And in defense, becoming an increasingly important potential market for us, there are a number of secular demand drivers going obviously here in the background.

Eben Upton: There are a number of secular demand drivers going on, obviously here in the background, a trend in the West towards rearmament, a recognition that we have probably wound down the size of our armed forces further than is wise. A growing recognition, driven probably by some of the experiences that are happening today in Ukraine, of the importance of autonomy, of advanced autonomy, of distributed decision-making, in mobile platforms. A pivot away, I think, from what we might call exquisite platforms, towards high cost, low volume platforms with very long development times, with development times on the order of decades, towards lower cost, attritable, lower cost consumable platforms.

Speaker #1: A trend in the West towards rearmament. A recognition that we have probably wound down the size of our armed forces further than is wise.

Speaker #1: A growing recognition, driven probably by some of the experiences that are happening today in Ukraine, of the importance of autonomy—of advanced autonomy, of distributed decision making—in mobile platforms.

Speaker #1: A pivot away, I think, from what we might call exquisite platforms towards high cost, low volume, platforms with very long development times, with development times on the order of decades, towards lower cost, attritable, lower cost consumable platforms.

Speaker #1: And enabling that transition, a move from only using military specification electronics to a growing focus on COPS, commercial off-the-shelf technology, as I say, as an enabler both for the cost and the availability and sustainability of these platforms.

Eben Upton: Enabling that transition, a move from only using military specification electronics to a growing focus on COTS, commercial off-the-shelf technology, as I say, as an enabler both for the cost and the availability and sustainability of these platforms. As we look more at the defense markets, really struck by the range of applications that Raspberry Pi fits very neatly into. Whether that is fixed hardware back at base, whether it is driving display screens or aggregating data in a fixed location. Whether it is deployed static hardware, so sensor platforms in the field, RFID goods in the field. Or whether it is providing intelligence and autonomy for mobile unmanned platforms like UAVs, USVs, and UGVs. All of these markets, just like our commercial customers, all of these markets stand to benefit from Raspberry Pi's distinctive value proposition.

Speaker #1: As we look more at the defense market, really struck by the range of applications that Raspberry Pi fits very neatly into. Whether that's fixed hardware back at base, whether it's driving display screens or aggregating data in a fixed location, whether they are whether it is deployed static hardware, so sensor platforms in the field, RFD coils in the field, or whether it's providing intelligence and autonomy for mobile unmanned platforms like UAVs, USVs, and UGVs.

Speaker #1: And all of these markets, just like our just like our commercial customers, all of these markets stand to benefit from Raspberry Pi's distinctive value proposition.

Speaker #1: Industry leading price performance, availability and long-term support, and the ability to manufacture these products and sustain them at scale. A word on silicon. Silicon really remains at the heart of Raspberry Pi's medium-term growth strategy.

Eben Upton: Industry-leading price performance, availability of long-term support, and the ability to manufacture these products, and sustain them at scale. A word on silicon. Silicon really remains at the heart of Raspberry Pi's medium-term growth strategy. It has been, and it continues to be an enabler for our single board computer and module business. We build silicon because it allows us to build better board level products, better modules, better SBCs, than we could build otherwise. But increasingly it is becoming a business for us in its own right. We saw lower sales. We saw slightly lower sales in H1 2026 than we did in the comparable period in 2025. It was a strong comparable period I think. We had a number of large OEM orders in that period. But then we followed, in July and August this year, absolutely fantastic months.

Speaker #1: It has been and it continues to be an enabler for our Singapore computer and module business. We build silicon because it allows us to build better board-level products, better modules, better SBCs, than we could build otherwise.

Eben Upton: We build silicon because it allows us to build better board level products, better modules, better SBCs, than we could build otherwise. But increasingly it is becoming a business for us in its own right. We saw lower sales. We saw slightly lower sales in H1 2026 than we did in the comparable period in 2025. It was a strong comparable period I think. We had a number of large OEM orders in that period. But then we followed, in July and August this year, absolutely fantastic months.

Speaker #1: But increasingly, it's becoming a business for us in our own right, in its own right. We saw lower sales. We saw slightly lower sales in the first half of 2026 than we did in the comparable period in 2025.

Speaker #1: It was a strong comparable period, I think. We had a number of large OEM orders in that period. But then we followed July and August this year, absolutely fantastic months.

Speaker #1: August in particular, a record month for sales of the RP2350 platform, the two-year-old second generation microcontroller platform. Particularly strong sales and particularly strong and particularly broad-based sales for those products in China.

Eben Upton: August in particular, a record month for sales of the RP2350 platform, the two-year-old second generation microcontroller platform. Particularly strong sales and particularly strong and particularly broad-based sales for those products in China. We continue to develop engagements with tier 1 and tier 2 OEMs in the West. That is inevitably a slower burn because of the Western engineering culture is inevitably a slower engineering culture. But we are confident that those will mature in due course into design wins. We have taped out in H2, in the last month. We have taped out what our next generation semiconductor product to get that back at the start of next year, and that may, in due course, lead to future board products, future innovative board products, and future standalone semiconductor products. Finally, a word on outlook. We expect the H2 volumes to be higher than H1 volumes.

Speaker #1: We continue to develop engagements with tier one and tier two OEMs in the West. That's inevitably a slower burn because of the Western engineering culture is inevitably a slower engineering culture.

Speaker #1: But we are confident that those will mature in due course into design wins. We have taped out in the second half, in the last month.

Eben Upton: We have taped out in H2, in the last month. We have taped out what our next generation semiconductor product to get that back at the start of next year, and that may, in due course, lead to future board products, future innovative board products, and future standalone semiconductor products. Finally, a word on outlook. We expect the H2 volumes to be higher than H1 volumes.

Speaker #1: We have taped out what our next-generation semiconductor product, and we expect to get that back at the start of next year. That may, in due course, lead to future board products, future innovative board products, and future standalone semiconductor products.

Speaker #1: And finally, a word on outlook. We expect second-half volumes to be higher than first-half volumes. What's going on there? Continued extremely strong demand.

Eben Upton: What is going on there? Continued extremely strong demand. The backlogs that we both referred to earlier are 2.6 million units at the end of H1. Those are large backlogs. They are probably verging on unhealthily large backlogs. So there is a great focus in the organization at the moment on reducing those backlogs, not to zero, but the aspiration is to get those backlogs back towards the levels they were at at the start of the year. To support that, we have been making a number of co-investments with our partners at Sony in order to increase their production rate, and we expect that Q4 will see the full benefit of all of those investments.

Speaker #1: The backlogs that we've both referred to earlier—2.6 million units at the end of the first half—those are large backlogs. They're probably verging on unhealthily large backlogs.

Speaker #1: So there's a great focus in the organization at the moment on reducing those backlogs—not to zero, but the aspiration is to get those backlogs back towards the levels they were at at the start of the year.

Speaker #1: To support that, we've been making a number of co-investments with our partners at Sony in order to increase their production rate, and we expect that Q4 will see the full benefit of all those investments.

Speaker #1: Despite this taper off, we saw some exceptional unit economics in the first half of the year, as some of the price increases that we've put through interacted with the low cost inventory that we were holding that we came into the year holding.

Eben Upton: Despite this taper off, we saw some exceptional unit economics in the H1 of the year as some of the price increases that we have put through interacted with the low-cost inventory that we were holding, that we came into the year holding. Despite the taper off of those exceptional unit economics, we do, as a result of those increased unit volumes, expect adjusted EBITDA to exceed market estimates somewhat. We are now holding enough DRAM. We have enough DRAM in our hands and on secure order to meet our requirements for the remainder of this year and into next year. We are excited about what 2027 and 2028 will bring for Raspberry Pi. I am now happy to take questions.

Speaker #1: Despite the taper off of those exceptional unit economics, we do, as a result of those increased unit volumes, expect adjusted EBITDA to exceed market estimates somewhat.

Speaker #1: We are now holding up around we have enough RAM in our hands and on secure order to meet our requirements for the remainder of this year.

Speaker #1: And into next year, and we're excited about what 2027 and 2028 will bring for Raspberry Pi. So I'm now happy to take questions. Fantastic Evan Richards.

Operator: Fantastic. Evan, Richard, thank you very much indeed for your presentation. Ladies and gentlemen, please do continue to submit your questions using the Q&A tab situated on the top right corner of your screen. While the company takes a few moments to review those questions submitted today, I would like to remind you that a recording of this presentation, along with a copy of the slides and the published Q&A, can be accessed via your investor dashboard. Evan, Richard, as you can see, we have received a number of questions throughout today's presentation. If I may now hand back to you and kindly ask you to read out the questions where appropriate to do so, and I will pick up for you both at the end. Thank you.

Speaker #1: Thank you very much indeed for your presentation. Ladies and gentlemen, please do continue to submit your questions using the Q&A tab situated at the top right corner of your screen.

Speaker #1: While the company take a few moments to review those questions submitted today, I would like to remind you that a recording of this presentation, along with a copy of the slides and the published Q&A can be accessed by our investor dashboard.

Speaker #1: I'm Evan Richards. As you can see, we have received a number of questions throughout today's presentation. If I may, I'll now hand back to you and kindly ask you to read out the questions, where appropriate to do so, and I'll pick up from you both at the end.

Speaker #1: Thank you.

Speaker #2: Okay, let's do it. So you're going to read them out, and I'm going to read them out. Mostly, I'm just trying to familiarize myself with the order they arrived in.

Eben Upton: Okay, let us do it. You are going to read them out?

Richard Boult: I am going to read them out.

Eben Upton: I am just trying to.

Richard Boult: Mostly I am going to familiarize myself with the order they arose in. What new products are in development? That is a good starter.

Speaker #2: What new products are in what new products are in development while I think I think we I think we share with another another fruit named company a tendency not necessarily to talk enormously about about what products we have in development.

Eben Upton: Well, I think we share with another fruit named company a tendency not necessarily to talk enormously about what products we have in development. I think we can say that it is uncontroversial to say that we obviously need to be starting to think about the next generation of our two platforms. We need to be thinking about the next generation of the big Raspberry Pi platform, the Linux-based platform, the thing which will eventually become Raspberry Pi 6, and the next generation of the small and microcontroller platforms. The thing which will generate both Raspberry Pi Pico 3 and the next generation of our P2 series microcontroller. Obviously, we are still some distance from both of those platforms are still fairly five and two platforms are still fairly new.

Speaker #2: I mean, I think we can say that, you know, it is it's uncontroversial to say that we obviously need to be starting to think about the next generation of our two platforms.

Speaker #2: So we need to be thinking about the next generation of the big Raspberry Pi platform—the Linux-based platform—which will eventually become Raspberry Pi 6.

Speaker #2: And the next generation of the small and microcontroller platforms, the thing which will generate both Raspberry Pi Pico 3 and the next generation of RP2 series microcontroller.

Speaker #2: Obviously, we are still some distance from both of those platforms are still fairly five and two platforms are still fairly new. But certainly, we're now reaching a point in their lifecycle where I think attention does turn to future products.

Eben Upton: But certainly we are now reaching a point in their life cycle where I think attention does turn to future products. I think we have said that you are likely to see more Raspberry Pi Pico 2 products. In the Raspberry Pi Pico generation, we have the Pico product and the Pico W product, the wireless and non-wireless product. I think it is likely you will see from us over the next 12 to 18 months, you are likely to see from us Pico 2 products which have a broader range of networking places. We have talked about that in the past.

Speaker #2: I think we've said that you're likely to see more Raspberry Pi Pico 2 products. In the Raspberry Pi Pico generation, we have the Pico product and the Pico W product—the wireless and non-wireless products.

Speaker #2: I think it's likely you will see from us over the next 12 to 18 months, you're likely to see from us Pico 2 products, which have a broader range of network interfaces.

Speaker #2: We've talked about that in the past, and pretty much all of those—you think about one of the big hitter accessory items: power supplies, displays, cameras—I think it's fairly likely that we will try to find ways to extend that.

Eben Upton: Pretty much all of those you think about, one of the big hits or accessory items, power supplies, displays, cameras. I think it is fairly likely that we will try to find ways to extend that as we get a better understanding of how our various customers, our enthusiast customers, our industrial and embedded customers. As we get a better idea of what they appreciate, what they like, and what they do not like about our existing offering in that space, I think you will see us broaden that offer out a little bit.

Speaker #2: As we get a better understanding of what of how our various customers are enthusiast customers, our industrial embedded customers, as we get a better idea of what they appreciate, what they like and what they don't like about our existing offering in that space, I think you'll see us kind of you'll see us broaden that broaden that offer out a little bit.

Speaker #3: One day, that we six generally follows five, doesn't it?

Richard Boult: One day there will be a 6 generally follows 5, does not it?

Speaker #2: Yes, usually.

Eben Upton: Yes, usually.

Speaker #3: I've got a couple of questions here.

Richard Boult: Got a couple of questions here. Congratulations.

Eben Upton: One plus one is two.

Speaker #2: One plus one is two.

Richard Boult: Yeah, thank you for that. Yes, I got that right, but I had to think hard.

Speaker #3: Yeah, thank you for that. Yes, I've got that right. I had to think hard.

Speaker #2: I'm going to bring my son to the next one.

Eben Upton: Can bring my son to the next one.

Richard Boult: There is a couple of questions in here about our relationship with Arm.

Speaker #3: There are a couple of questions in here about our relationship with all.

Speaker #2: Yes.

Eben Upton: Yes.

Speaker #3: How has that relationship been developing? What's it like having them as a shareholder? I mean.

Richard Boult: How has that relationship been developing? What is it like having them as a shareholder?

Eben Upton: Arm have been an enormously supportive shareholder. Obviously, they were a pre-IPO investor. They cornerstone the IPO. They bought a certain amount of the Foundation's

Speaker #2: I mean, they've been an enormous I'm have been enormously supportive shareholder, obviously. They they were They cornerstoned the IPO. They they bought a certain amount of the of the foundations the foundation salt block in April and aren't participating in that secondary sale.

Richard Boult: Yeah

Eben Upton: Foundation sold block in April.

Richard Boult: Yeah, sure.

Eben Upton: Arm participated in that secondary sale. So been a very supportive shareholder. I often think of this, the substance of the questions is, what has having Arm as a shareholder done to the strategic relationship? I think I have always viewed this the other way around. I have always seen this as the shareholding, recognizing, validating, being a consequence of an existing strategic relationship. Every Raspberry Pi is an Arm computer.

Speaker #2: So they've been a very supportive shareholder. I think I often think of this, you know, the sort of the substance of the questions is, what what has having ARM as a shareholder done to the strategic relationship?

Speaker #2: I think I've always viewed this the other way around. I've always seen this as the shareholding recognizing, validating being a consequence of an existing strategic relationship.

Speaker #2: You know, we have been every Raspberry Pi is an ARM computer. It's not a huge number of ARM PCs. I as a child had the child of the 1980s have great great affection.

Richard Boult: Yeah.

Eben Upton: We sold a huge number of Arm PCs since childhood. Childhood of the 1980s have great affection. The Acorn Archimedes have a great affection for the Arm architecture since childhood. So this is something which has always been important to us. I think that having the opportunity to have Arm as a shareholder has been very helpful for us. It has deepened the level, I think, from which we can have these strategic discussions with them. There will be many more Arm-based Raspberry Pi products in the future. I think they are excited about it. We see the roadmap, we see the products they have announced. We are excited about-

Speaker #2: They call it Archimedes. I have a great affection for the ARM architecture since childhood. So this is something that has always been important to us.

Speaker #2: I think that having the opportunity to have ARM as a shareholder has been very helpful for us. It has deepened the level, I think, on which we can have these strategic discussions with them.

Speaker #2: There will be many more ARM-based Raspberry Pi products in the future. I think they're excited about it. We see the roadmap; we see the products they've announced.

Speaker #2: We're excited about.

Speaker #3: They do, and they engage at all levels, including senior levels within ARM as well. It's great.

Richard Boult: They do engage at all levels.

Eben Upton: Yeah.

Richard Boult: Senior levels within Arm as well. It is very interesting.

Speaker #2: Yeah, we get great support from them. We're excited about the core roadmap. We're excited about the technology roadmap. We're excited about finding ways to integrate that roadmap with our own roadmap to keep building more high-performance ARM-based products.

Eben Upton: Yeah, we get great support from them. We are excited about the core roadmap. We are excited about the technology roadmap. We are excited about finding ways to integrate that roadmap with our own roadmap to keep building more high-performance Arm-based products. We are a very happy Arm partner.

Speaker #2: Very very very we are a very happy we are a we are very happy ARM partner. Very happy ARM licensee. Very happy ARM partner.

Richard Boult: Yes.

Eben Upton: Very happy Arm licensee, very happy Arm partner. The Arm partner meeting in August is an absolute highlight.

Speaker #2: The ARM partner meeting in August is an absolute highlight.

Speaker #3: As they were, being to you when you released a risk protocol.

Richard Boult: They will not meet you when you release the Raspberry Pi 5.

Speaker #2: Indeed.

Eben Upton: Indeed.

Richard Boult: What steps have been taken to ensure we remain a UK public company? As a UK shareholder, it would be extremely disappointing to see us taken over by a foreign entity.

Speaker #3: What steps are being taken to ensure we remain a UK public company? As a UK shareholder, it's extremely disappointing to see us potentially taken over by a foreign entity.

Speaker #2: I mean, I'm not sure that I'm not sure there is an enormous amount amount there are an enormous number of specific things that one can do here.

Eben Upton: I'm not sure. I'm not sure there is an enormous amount. There are an enormous number of specific things that one can do here other than we do our best. We are very proud.

Speaker #2: Other than, you know, we do our best. We are very proud.

Speaker #3: We we try to be as amusing and entertaining in meetings like this to satisfy our UK shareholders that they continue to want us to be independent rather than the.

Richard Boult: We try to be as amusing and entertaining in meetings like this to satisfy our UK shareholders that they continue to want us to be independent rather than

Speaker #2: So we are a look, we are a proud we're a proud UK listed. Public company. We chose to list in the UK. We you know, we we design our products in the UK.

Eben Upton: Look, we are a proud UK-listed public company.

Richard Boult: Yeah.

Eben Upton: We chose to list in the UK. We design our products in the UK, manufacture our products in the UK. We are listed in the UK. I don't think we do any of these things for patriotic reasons. We do these things because we believe that the UK is the best place to do each of those three things. We've had a fantastic experience, I think, as a UK PLC. Our aspiration, what are we trying to grow here? We are trying to build a globally meaningful business from the UK, based in the UK, listed in the UK, manufacturing in the UK. While we have the opportunity to, we will continue to do that.

Speaker #2: We manufacture our products in the UK. We are listed in the UK. We don't do any of these things for I don't think we do any of these things for patriotic reasons.

Speaker #2: We do these things because we believe that the UK is the best place to do each of those three things. So, you know, we've had a fantastic experience, I think, as a UK PLC.

Speaker #2: Yeah, our aspiration—what are we trying to grow here? We are trying to build a globally meaningful business from the UK: based in the UK, listed in the UK, manufacturing in the UK.

Speaker #2: And, you know, while we have the opportunity to, we will continue to do that.

Speaker #3: Yeah.

Richard Boult: Yeah.

Speaker #2: Do we have a CRA question? Yeah, let's do the cyber resilience that question. So so this is so this is sort of this is a couple of questions about the the EU cyber resilience act.

Eben Upton: Do we have a CRA question?

Richard Boult: Yes. Would you like Okay.

Eben Upton: Yeah, let's do the Cyber Resilience Act question. There's a couple of questions about the EU Cyber Resilience Act and the requirement for full compliance in December next year. I think this is something that we've seen coming for a long time.

Speaker #2: And the requirement for full compliance in December next year. I think this is this is something that we have seen something we've seen coming for a long time.

Richard Boult: Yeah.

Speaker #2: It is it drove over the kind of five years, I guess over the period that we were designing Raspberry Pi 5, the awareness that there was going to be increasing government action and increasing government awareness around IoT security drove the incorporation into the Raspberry Pi 5 hardware platform and the Raspberry Pi the RP2350, the second generation microcontroller drove the incorporation of hardware features of this pretty clear what hardware features would be required in order to allow to give people at least the potential to build compliant solutions.

Eben Upton: It drove over the kind of five years, I guess, over the period that we were designing Raspberry Pi 5, the awareness that there was going to be increasing government action and increasing government awareness around IoT security drove the incorporation into the Raspberry Pi 5 hardware platform and the Raspberry Pi, the RP2350, the second-generation microcontroller, drove the incorporation of hardware features. It was pretty clear what hardware features would be required in order to allow, to give people at least the potential.

Richard Boult: Yeah

Eben Upton: to build compliant solutions. There is then a software layer on top of that. We had largely by, I guess, two years ago when we launched RP2350, we had largely completed the process of delivering the underpinnings to CRA compliance. What you have seen subsequently is us then building out both the software level required, particularly some of these for secure updates, secure firmware updates on our platform. But also then the tools that people need in order to build the regulatory artifacts, certification artifacts that sit alongside. It is no longer enough to build a firmware image.

Speaker #2: There's then a software layer on top of that. So we we have largely by, I guess, two years ago when we launched 2350, we've largely completed the process of delivering the underpinnings for CRA compliance.

Speaker #2: What you've seen subsequently is us then building out both the software level required, particularly some of these for secure updates, for secure firmware updates on a platform.

Speaker #2: But also, then, the tools that people need in order to build the regulatory artifacts, certification artifacts that sit alongside. So it's no longer enough to build firmware.

Richard Boult: Yes, I think the first bit we did, because it is the right thing to do.

Speaker #3: Yes. I think the first bit we did because it's the right thing to do.

Speaker #2: Yes, that's right. And you no longer it's no longer sufficient to simply build the firmware image. You have to create the certification, the regulatory artifacts therefore.

Eben Upton: Yes, that is right. It is no longer sufficient to simply build the firmware image. You have to create the certification, the regulatory artifacts, the SBOM. You have to be able to generate the SBOM as well. A lot of those investments that we made in automating, systematizing the process of building the software image for a Raspberry Pi IoT device, the design of those systems incorporated the knowledge that people were going to have to conform to the CRA. I think that has gone pretty well. In terms of its impact on us as a business, it is driving, we already see it driving a tendency to use our software, not just to buy our hardware devices, but to use our software and to use our systems.

Speaker #2: You have to be able to generate the SBOM as well. And so a lot of those investments that we made in automating and systematizing the process of building the software image for a Raspberry Pi OT device were—the design of those systems incorporated the knowledge that people were going to have to conform to the CRAs.

Speaker #2: And I think that's gone pretty well in terms of its impact on us as a business. It is driving—we already see it driving—a tendency to use our software, not just to buy our hardware devices, but to use our software and to use our systems when you're assembling the firmware.

Richard Boult: Yeah

Eben Upton: when you are assembling the firmware. That is extremely attractive to us because it increases the stickiness of the platform. It makes it more salient to our customers. It increases the stickiness of the platform. Once you have designed your Raspberry Pi and that is working well, it increases the switching cost away from Raspberry Pi as an endpoint. I think this is something where, like many of these things, you see a difficult thing, you see a challenge, and at first you are sad that you are going to have to do some work. Then after a little while, you realize that having done that work, that work becomes a plank, another plank in your story about competitive advantage.

Speaker #2: And that's extremely attractive for us because it increases the stickiness of the platform. It makes the platform more saleable to our customers. It increases the stickiness of the platform once you've designed Raspberry Pi and that's working well.

Speaker #2: It increases the switching cost away from Raspberry Pi as a vendor. So I think this is something where, you know, like many of these things, you see a difficult thing.

Speaker #2: You see a challenge. And at first you are sad that you're going to have to do some work. And then after a little while, you realize that having done that work, that work becomes a plank, another plank in your story about competitive advantage.

Speaker #3: Yes, yes. You’re lying outside a few times with investors. If you want to buy shares in Raspberry Pi, spend some money on buying a competitor’s product first and just appreciate the quality of what we’ve made.

Richard Boult: Yes. Yes. The line I have heard a few times with investors of, "If you want to buy shares in Raspberry Pi, spend some money on buying a competitor's product first and just appreciate the quality of what we have made.

Speaker #2: The other question was about the UK government. We expect—and whether we are engaged with the UK government on this, particularly around ensuring that whatever legislation is passed permits us to continue to adopt open source software development methods.

Eben Upton: The other question was about the UK government. We expect, and whether we are engaged with the UK government on this, particularly around ensuring that whatever legislation is passed commits you to continue to adopt open source software development methods. We are not specifically engaged on this. We do expect the UK government to largely be a follower here. We do not expect the government here in the UK to diverge meaningfully from continental European practice. We believe that it is unlikely that the European version of this will evolve in ways which are inimical to people's ability to use open source software development methods. I think there is a broad consensus actually, that open source approaches are the approaches.

Speaker #2: We are not specifically engaged on this. We do expect the UK government to largely be a follower here. We don't expect the government here in the UK to diverge meaningfully from continental European practice.

Eben Upton: We do not expect the government here in the UK to diverge meaningfully from continental European practice. We believe that it is unlikely that the European version of this will evolve in ways which are inimical to people's ability to use open source software development methods. I think there is a broad consensus actually, that open source approaches are the approaches. Particularly actually open source approaches allied to machine learning tools, allied to AI tools, are the way to build systems which have that level of robustness and usability at deployment.

Speaker #2: And we believe that it is unlikely that the European approach to this will evolve in ways which are inimical to people's ability to use open source software development methods.

Speaker #2: I think there is a broad consensus, actually, that open source approaches are the approaches—particularly, actually, open source approaches allied to machine learning tools and AI tools are the way to build systems which have that level of robustness that you'd be happy to deploy.

Eben Upton: Particularly actually open source approaches allied to machine learning tools, allied to AI tools, are the way to build systems which have that level of robustness and usability at deployment.

Speaker #3: Changing gear. What's our expectation for unit sales for the semiconductor business this year and in the medium term? How quickly should we expect this business to grow?

Richard Boult: Changing gear. What is our expectation for unit sales of semiconductor business this year and in the medium term? How quickly should we expect this business to grow? I think in the short term, we are talking microcontroller sales. We would be very sad if it does not start with a one-

Speaker #3: I mean, think in the short term—we're talking microcontroller sales. You'd be very sad if it doesn't start with a '1.'

Speaker #2: Yeah. Yeah.

Eben Upton: Yeah

Speaker #3: This year, so probably in the sort of 10 to 12.

Richard Boult: this year. So probably in the sort of 10 to 12 scale.

Speaker #2: Yeah. I think I think the I think the the sort of the you can put a box around it, which is this is a let's say on the order of 30 billion unit just by.

Eben Upton: I think you can put a box around it, which is this is a, let's say, on the order of 30 million unit, just my

Richard Boult: Yeah

Speaker #2: On the order of 30 billion unit business. We haven't aspiration. There are it's an oligopoly. It's a it's a five player oligopoly largely. So roughly 75 to 80% of that goes to five vendors that I probably can't I can't name four of them either today.

Eben Upton: on the order of a 30 million unit business. We have an aspiration. It is an oligopoly. It is a five-player oligopoly largely. So roughly 75% to 80% of that goes to five vendors who I probably cannot name. I can probably name four of them blind during the day. But the remainder of that market is actually quite fragmented.

Speaker #2: But the remainder of that market is actually quite fine with it. And so an aspiration to go and own one or two percent of that market is not outlandish at all.

Richard Boult: Yeah.

Eben Upton: An aspiration to go and own more than 2% of that market is not outlandish at all.

Speaker #2: So I think we have an aspiration to grow the business out into what’s going to be 100 percent units. I think that getting into double-digit millions is an important step along the road.

Richard Boult: Yes.

Eben Upton: I think we have an aspiration to grow the business out into the hundreds of millions of units. I think getting into double-digit millions is an important step along the road. We still have to sustain. We have seen a great growth rate in these products since we launched them five years ago.

Speaker #2: Obviously, we still have to sustain. We've seen a great growth rate in these products since we launched them five years ago. I think we have to sustain that growth rate, and obviously, it becomes easier and harder to sustain big growth as you get bigger. It becomes easier and harder to sustain a large growth rate.

Richard Boult: Yeah.

Eben Upton: I think we have to sustain that growth rate. It becomes easier and harder to sustain big growth. As you get bigger, it becomes easier and harder to sustain large growth rates. But I think there is an aspiration to do that. Where do we want to be? I think we have said very consistently, a decade after the IPO, the aspiration is to have grown the traditional business, the boards and modules business significantly, but for semiconductors to have come up alongside that, come up underneath that to a point where they are roughly coequal. Some of that heavy lifting I think will be done by ASP. I am sure some of the products introduced in the future will have higher ASPs, but a lot of that has to be done by volume.

Speaker #2: But I think there is an aspiration to do that. Where do we want to be? I think we've said very consistently a decade after the IPO, the aspiration is to have grown the the traditional business, the the boards and modules business significantly and but for semiconductors to have come up alongside that come up underneath that to a point where they're roughly equitable.

Speaker #2: Now, some of that heavy lifting I think will be done by ASP. I'm sure you know some of the products they're producing in the future will have higher ASPs.

Speaker #2: But a lot of that has to be done by volume. So I think there are sort of two directions in which we are saying the business really needs to find a way into at least the hundreds of millions of units a year of silicon in order to meet those — in order to feel that we are, you know, these are exceptional pieces of hardware.

Eben Upton: I think there are two directions in which we are saying the business really needs to find a way into at least the hundreds of millions of units a year of silicon in order to meet those, in order to feel that we are. These are exceptional pieces of hardware. They really are. I think connoisseurs, we have heard and seen them described sometimes as connoisseurs microcontrollers. The better an engineer is, the more attractive they are to the Raspberry Pi silicon platform. So I think we have something that is very compelling. I think if we are not able to get into that regime, then I would personally feel that we have somehow taken a great product and somehow. We will not have failed technically. We will have failed in translating that technical excellence into market revenue.

Speaker #2: They really are. They are, I think, commissars. We've heard, seen them, seen them described sometimes as commissars, microcontrollers. The better an engineer is, the more attractive they are.

Speaker #2: To the Raspberry Pi silicon platform. So I think we have something that's very compelling. I think if we aren't able to get into that regime then I, I would personally feel that we've somehow taken a great product and hollowed out its shores, that we somehow won't have failed technically.

Speaker #2: We will have failed to translate that technical excellence into market achievement.

Speaker #3: Yeah, yeah. Quick one — so, reference market consensus for adjusted EBITDA for the full year. I think we're seeing numbers in the sort of range of low sixties, sort of $60 to $65 million.

Richard Boult: Yeah. A quick one from. Reference market consensus for adjusted EBITDA for the full year. I think we are seeing numbers in the range of low 60s, 60, 65, I think across the spread of analysts that we talk to.

Speaker #3: I think across the spread of analysts that we talk to.

Speaker #2: I think when we reference market consensus there, that is market consensus as of yesterday evening. So, I think market consensus may have moved this morning.

Eben Upton: I think where we reference market consensus there, that is market consensus as of yesterday evening. I think market consensus may have moved this morning. This is market. I think we were conveying a belief that that prior consensus was a little conservative relative to how we see the business performing in the full year.

Speaker #2: So, this is market. You know, I think we're conveying a belief that the prior consensus was a little conservative, relative to how we see the business performing in the full year.

Speaker #3: When do you expect to make a positive free cash flow? It is not something that is alien to us as an organization. I mean, I think from 2012 through to through even after I arrived, we were generating free cash flow.

Richard Boult: When do you expect to make a positive free cash flow? It is not something that is alien to us as an organization. I think from 2012 through to even after I arrived, we were generating free cash flow.

Eben Upton: It wasn't just you.

Speaker #3: It wasn't just me. I think over the last few years, we have invested significantly in CapEx, but more importantly, we have ridden through a very turbulent supply chain period.

Richard Boult: It wasn't just me. We have, I think over the last few years, we have invested significantly in CapEx, but more importantly, we have ridden through a very turbulent supply chain period. We've taken that opportunity to use our balance sheet. Yes, we've ended up with negative free cash flow. I think the answer to this question is probably when do we think the turbulence and crisis over memory is going to stop? If we believe that is in 2028, which I think is probably now the earliest point, ideally when additional fab capacity overwhelms the demand from data centers, that is a period when we should see some scope to start to reduce inventory levels or at least hold them steady while the business is bigger.

Speaker #3: We've taken that opportunity to use our balance sheet. Yes, we've ended up with negative free cash flow. I think the answer to this question is probably, when do we think the turbulence and crisis over memory is going to stop?

Speaker #3: If we believe that is in 2028—which I think is probably now the earliest point, ideally—when additional fab capacity overwhelms the demand from data centers, that is a period when we should see some scope to start to reduce inventory levels, or at least hold them steady while the business is bigger.

Speaker #2: I think you're you're in a you're in a situation at the moment where you are you're holding quite a lot of DRAM.

Eben Upton: I think you are in a situation at the moment where you are holding quite a lot of DRAM because you want to make sure that in the event of an availability shock, you are able to make a product.

Speaker #3: Yeah.

Speaker #2: Because you want to make sure that in the event of an availability shock. You are able to make a product. And at the same time holding that DRAM is expensive because DRAM is expensive to fix.

Richard Boult: Yeah.

Eben Upton: At the same time, holding that DRAM is expensive because DRAM is expensive to place.

Richard Boult: Yes.

Speaker #2: I think at the point where the market, where the cycle flips over, I don't think you're necessarily going to see us—we're holding it, so let's say between three and four months of DRAM, DRAM for the moment.

Eben Upton: I think at the point where the market, when the cycle flips over, I do not think you are necessarily going to see us withholding it, let us say, between 3 and 4 months of DRAM. I think if the market flips over, you are not going to see us holding 3 or 4 months of DRAM, and even if we were holding 3 or 4 months of DRAM, that would be much less cash consuming. I suspect we would be generating positive. I know that we would be generating positive free cash flow now if we had not had a memory price shock.

Speaker #2: I think if the if the if the the market flips over you're not going to see us holding three or four months of of DRAM and even if we were holding three or four months of DRAM that would be much less cash consuming.

Speaker #2: So I think it is—had we not—I suspect we would be generating positive, I know that we would be generating positive free cash flow now if we had not had a memory price shock.

Speaker #3: Yeah.

Richard Boult: Yeah.

Eben Upton: I am pretty confident that once we get out the other end of this, the business will revert to what I still believe, having run the business since 2012, what I still believe is the natural behavior of Raspberry Pi, that it is a fairly cash generative business.

Speaker #2: I'm pretty confident that once we get out the other end of this the business will revert. What I still believe having run the business since 2012 what I still believe is the natural behavior of Raspberry Pi that it is a fairly cash generative business.

Speaker #3: Yeah, okay. Probably quick on margins—have been improving lately, but are below the 2021 level. Is the plan to return to those levels, or is that unrealistic?

Richard Boult: Okay. Margins have been improving lately but are below the 2021 level. Is the plan to return to those levels, or is that unrealistic? I think margin here we are talking about gross profit as a percentage of revenue. That is something that I think we have talked about before. Our revenue number does include a number of different lines with really quite different margin structures. The royalty income that was, I think in 2021, 75% of our unit volume was going through Farnell at near 100% margin. Whereas now 75%, 80% is direct, so the Farnell licensee piece is only 20%. I think it will be some time before we think about going back to a 2021 level. The way that we view the business, certainly I do, is based on that gross profit per unit. What dollars are we making for each board shipped?

Speaker #3: I think when margin here we're talking about gross profit as a percentage of revenue that is something that I think we've talked about before.

Speaker #3: Our revenue number does include a number of different lines with really quite different margin structures. You know, the royalty income that was, I think, in 2021, 75% of our unit volume was going through Farnell at near 100% margin.

Richard Boult: The royalty income that was, I think in 2021, 75% of our unit volume was going through Farnell at near 100% margin. Whereas now 75%, 80% is direct, so the Farnell licensee piece is only 20%. I think it will be some time before we think about going back to a 2021 level. The way that we view the business, certainly I do, is based on that gross profit per unit. What dollars are we making for each board shipped?

Speaker #3: Whereas now, 75–80% is direct, so the Farnell licensee piece is only 20%. So I think it will be some time before we think about coming back to a 2021 business. Certainly, what I do is based on that gross profit per unit—what dollars are we making for each board shipped. That allows us to stand back from which route it's going to market and really focus on what profit it makes, rather than focusing on a percentage.

Richard Boult: That allows us to stand back from which route is it going to market and really focus on what profit does it make rather than focusing on a percentage. I think if we had tried to have maintained a margin percentage in this most recent year, if we continued to press for that with prices going up so much, yes, we would have a temporary boost of profit, but I think we would have opened up some quite frightening opportunities for competitors, particularly in the Far East, to come and take away our very strong presence across all customers.

Speaker #3: I think if we tried to maintain a margin percentage in this most recent year—if we continue to press for that, with prices going up so much—we would have increased our, yes, we would have had a temporary boost of profit, but I think we would have opened up some quite frightening opportunities for competitors, particularly in the Far East, to come and take away our very strong presence across all customers.

Speaker #2: I think it would have been a breach of faith with our OEM customers. I referred to a couple of very loyal customers there.

Eben Upton: It would have been a breach of faith with our-

Richard Boult: Yeah

Eben Upton: OEM customers. I referred to a couple of very loyal customers there during the presentation. When somebody designs Raspberry Pi into their product, even without the CRA sort of enhancements, they are incredibly sticky sockets. We are constantly alive to any perception that we are exploiting the stickiness of those sockets in order to extract value from our OEM customers. The decision to pass through DRAM costs rather than to margin and passing them-

Speaker #2: During a presentation, when somebody designs Raspberry Pi into their product, even without the CRA sort of enhancements, it is an incredibly, incredibly sticky socket.

Speaker #2: You know, we are constantly alive to any perception that we are exploiting the stickiness of those sockets in order to extract value from our OEM customers.

Speaker #2: The decision to pass through DRAM cost, rather than to margin, and pass DRAM cost through both ourselves and then our channel structure, has been why it's one of the most challenging decisions.

Richard Boult: Yeah

Eben Upton: a DRAM cost through both ourselves and our channel structure has been one of the most challenging decisions

Richard Boult: Yeah

Speaker #2: That I've had to make. In in my time at Raspberry Pi. I think it has been the right decision. It does have this poten it does have a it will have a negative effect on percentage margin but as Richard says when we manage the business we manage the business too.

Eben Upton: that I've had to make

Richard Boult: Yeah

Eben Upton: in my time at Raspberry Pi. I think it has been the right decision. It will have a negative effect on percentage margin. But as Richard says, when we manage the business, we manage the business to

Richard Boult: Yeah

Speaker #2: Units: What do I look at every day? Units. I look at gross profit per unit—it's that decomposition of overall gross profit that has my attention every day.

Eben Upton: units. What do I look at every day? Units. I'm looking for a profit, but it's that decomposition of overall gross profit that has my attention every day.

Speaker #3: Yeah. And I think we've talked about that over many years that it's remarkable that the business that you've you've built and talked about across the world that we've not seen a significant assault on this business by low cost competitors in China.

Richard Boult: Yeah. I think we've talked about that over many years, that it's remarkable the business that you've built and talked about across the world, that we've not seen a significant assault on this business by low-cost competitors in China, because they're not lower cost.

Speaker #2: And the and the low.

Speaker #3: And the lower cost.

Speaker #2: And a lot of that is about that humble approach.

Eben Upton: A lot of that is about that humble approach.

Speaker #3: Yeah.

Richard Boult: Yeah.

Speaker #1: That's great, guys. Sorry to interject there. We are coming up to the hour, so Ewan, if I may, just ask you for some closing comments to wrap up.

Operator: That's great, guys. Sorry to interject there. We are coming up to the hour. So Eben, if I may just ask you for some closing comments to wrap up.

Speaker #2: Thank you. This has been a spectacular half. I think we have had, you know, as I said, we are—we're proud to be a UK-listed company.

Eben Upton: Thank you. This has been a spectacular half. I think it has been we have had, as I said, we're a proud UK-listed company. We've had a good experience as a PLC. This has been by far the best half of that 2.5 years. This has been by far the best period. It's been wonderful to demonstrate the business's resilience during what is an extremely challenging time. More than anything else, it has been extremely satisfying to have been able to be of service to our customers, our enthusiast customers, our embedded customers, our industrial customers, to have been able to keep our product in production, to have been able to use some of the strategic memory inventory to buffer and minimize and delay the impact of some of these cost structure changes on the platform.

Speaker #2: We've had a good experience as a PLC. This has been by far the best half of that two and a little bit years.

Speaker #2: This has been, by far, the best period. It's been wonderful to demonstrate the business's resilience during what is an extremely challenging time. Over more than anything else, it has been extremely satisfying to have been able to be of service to our customers.

Speaker #2: Our enthusiast customers, our embedded customers, our industrial customers, too, have been able to keep our product in production; to have been able to use some of the strategic memory inventory to buffer, minimize, and delay the impact of some of these cost structure changes on the platform.

Speaker #2: You know, we hope that people understand the effort that's gone in in the organization. And so, last words—I'd like to take the opportunity, as I said, this is Richard's last—this is Richard's last set of Raspberry Pi results.

Eben Upton: We hope that people understand the effort that's gone into the organization. Last words, I'd like to take the opportunity, as I said, this is Richard's last set of Raspberry Pi results. He's been with us since 2019. He was our first full-time true finance officer and really a measure of the extent to which the business has evolved over the last seven years. I would just really very much like to thank Richard for his service and his friendship as we took the business through just quite the most remarkable period of evolution. So thank you.

Speaker #2: He's been with us since 2019. He was our first full-time Chief Financial Officer and really a measure of the extent to which this business has evolved over the last seven years.

Speaker #2: So, I would just really very much like to take this opportunity to thank Richard for his excellent first friendship as we took the business through just quite the most remarkable period of evolution.

Speaker #2: So thank you.

Richard Boult: Thank you.

Speaker #3: Thank you. I won't miss you too.

Eben Upton: Over to you two.

Speaker #1: That's great, guys. Thank you very much once again for updating investors today. I'm going to please ask investors not to close the session. As you know, we are automatically redirected to provider feedback, which will help the company better understand your views and expectations.

Operator: That's great, guys. Thank you very much once again for updating investors today. Could I please ask investors not to close this session, as you'll now be automatically redirected to provide your feedback, which will help the company better understand your views and expectations. On behalf of the management team, we would like to thank you for attending today's presentation, and good afternoon to you all.

Browse all earnings call transcripts

Q2 2026 Raspberry Pi Holdings PLC Earnings Call

Request TrialDemo

Raspberry Pi Holdings

Earnings