Play the conversation by clicking the orange play button above, or subscribe to the NextMarket podcast at iTunes, Stitcher Radio or RSS.
I had an opportunity to talk with Eben Upton, Founder of the Raspberry Pi Foundation.
For those of you not familiar with Raspberry Pi, it was conceived by Upton in 2006 as a low-cost basic computer that would sell for the price of a text book ($25) and enable kids from nearly any socioeconomic background to have a computer.
The impetus for the project was the precipitous dropoff in the quality of computer knowledge among incoming computer science applicants that Upton and his colleagues saw at the University of Cambridge's computer laboratory.
Fast-forward six or so years, and the Raspberry Pi has shipped 1 million units and is selling at a volume of a couple hundred thousand a month. I decided to talk to Eben about what's next for the foundation and what he believes it's legacy will be.
(UPDATE February 15th: We have included a transcript of the conversation. Transcripts are best-effort, so if you notice error, please let us know and we will fix!)
TRANSCRIPT OF CONVERSATION ON FEBRUARY 8TH, 2013: MICHAEL WOLF, EBEN UPTON
(Wolf) Well I have with me Ebon Upton, the man who really got everything going with Raspberry Pi. I think most people in the tech world have at least heard of it, but how would you explain Raspberry Pi, Ebon?
(Upton) Raspberry Pi is a credit card sized $20.00 to $35.00 computer, depending on the version, it runs Linux and it’s produced out of the Raspberry Pi foundation out of Cambridge in the UK. We created it in the hope that we could get it into children bedrooms and help them learn to program in the same way that those of us in the foundation learned how to program with our microcomputers in the 1980s.
(Wolf) Yeah and you really pointed to the influence with what the BBC did the 80s and you conceived of this in the 2005, 2006 time frame is that about right?
(Upton) Yeah that’s right. This came out of a realization at the University of Cambridge that we were seeing every year fewer and fewer people applying to study computer science and every year the things those people knew how to do got less impressive. We feel that the 1980s 8-bit micro computer were responsible having this wonderful line of power so for us this is an attempt to kind of reboot that pipeline.
(Wolf) You weren’t necessarily focused on emerging markets, but with the price point there has been interest. From the beginning did you conceive of a $25.00 price point, was that your target initially?
(Upton) Yes, the $25.00 came before pretty much everything else, $25.00 is supposed to the price of a textbook. I think it turns out that we were pretty confused about what textbooks cost. That was how we picked our price point, we’ve got to ask children to go out and buy this thing and you can’t make it too expensive if you want every child to be able to use. You can make it expensive but then have horrific access issues, and if you want everyone to be able to program on a computer you need make the thing affordable
(Wolf) Somewhere along the line you got momentum and it really started moving along. Was there any point in the five years from conception to February 2012 when you started shipping the first units that you doubted this thing would happen?
(Upton) We went through peaks and troughs, we did try a bunch of different roots because of the aggressive price point, and we ended up with prototypes that for one reason or another were inadequate. It was really only the availability of this Broadcom chip, which became available in 2009-2010 that made it feasible to do this. Until then we had concerns as to whether it was doable at all, even after we had some concerns as to whether we could really mass-produce it at the quoted price. In the end it all worked out, but there were continuous periods of six months when we thought: “Oh no, we are doomed,” but then later on that week you would look at the cost of the SD card connector. And you would say, “You know, the cost of the SD card is 2 % of the cost of the device and we’ve got to get that down” and we ended up with an SD card that costs pennies, but all of these things. You have a device that has sixty, eighty, a hundred different components on it, and each of those chips away at your $25.00. Once you’ve negotiated a good price for your ram, your processor, then there’s nothing really big left, and it’s just the vast number of little things, and the number of hours of supplier negotiation, not just from us but also from our partners.
(Wolf) And you’ve pointed to the Broadcom chip as being the central factor that allowed you to get to that price point. You actually went to work for Broadcom, how did that happen?
(Upton) So I actually went to work for Broadcom in 2006, and I pretty much quit academia and went to work in industry, because I think there is a lot of really good work that is done in the industry, a lot of fundamental engineering. One of the first things I realized was that these guys were making chips for mobile phones and they were selling for the same price as the micro controllers we were trying to work with, but they are a hundred times powerful. We tried early versions of the Broadcom chips, and in the end it was just about keeping a weather eye on what was coming down the pipeline. This chip came along when I was on the design team for the chip that we later used. And you wake up one and realize, you’re not just designing a great chip, but you’re also designing a chip for the side project that you were planning.
(Wolf) Initially production was in Asia or China and then you moved it to England, and the initial run was 10,000 or so. How did that sell out and why did you guys decide to move production to England?
(Upton) We made 10,000 as our first batch we thought that would last us for a bit. We sold 100,000 and took a 100,000 orders in the first day. So 10,000 lasted about a minute. The limit was really the ability of the websites to stay up, it was the sort of volume you would have for popular rock concerts so that was nice. In terms of moving back to the UK, we never saw any reason why you couldn’t build them in the UK, but everywhere we went we got these quotes which were just massive, enormously times the component cost and we weren’t able to do it. In the end we did what everyone does, which is to fold and go for the default option and build in Shenzhen like everyone else. Then we were approached by somebody who works at Sony in South Wales, who said, “we could build this, or at least judging by your publicly released data, we could build this.” So we gave them a little bit of data, really just the component count and a sketch of what the board was like, and they still said they could built it. So we then gave them all the manufacturing data, the thing you need to make a Pi: a description of how the boards stack up and the component positioning, and then came back to us after two weeks and said “We still think we can do it.” The Pi foundation is just the IP licensing organization we just own the design of the board, we own the trademark, and we license that to RS Component and Premier Farnell who then make them and distribute them. We introduced our two partners to Sony, and they then did a large amount of work with support from the foundation to do the production and engineering and then in August, it went into production.
(Wolf) Is it unusual to act like a contract manufacturer, it doesn’t seem like a model they traditionally do. Were they doing it because there is a bigger point to Raspberry Pi versus just making money?
(Upton) It turns out Sony does contract manufacturing. It’s a wonderful pitch you get access to the Sony supply chain, to the Sony quality, and the Japanese total quality management thing going on, and you can tap into that as a third party. This is the factory where they built the broadcast cameras; Sony’s extremely expensive HD broadcast cameras 60 to 70 percent of the market.
(Wolf) That’s a huge spectrum shift in term of price point.
(Upton) They use to build televisions there; it was originally their CRT television factory in Europe. They were building ten thousand CRTs a day, and that’s an incredible number of televisions. That business went away of course with the LCD and they started building these much lower volume but much higher value camera products and then really the Pi was a return on some level for them to making thousands of units a day.
(Wolf) Where are you in terms of production today, in terms of unit volumes?
(Upton) Across both sites we’re probably doing between 150,000-200,000 a month.
(Wolf) In terms of units, or dollars?
(Upton) In terms of units.
(Wolf) That’s actually a lot higher number than has been talked about before, isn’t it?
(Upton) We’ve been open for ten months, shipping for eleven months now. We’ve shipped about a million units so it’s about 100,000 a month of average volume, but of course that includes a backload because it took us some time to ramp up at the start so the first few months we were making 10,000 a month.
(Wolf) And how many Raspberry Pis are out there at this point, do you know?
(Upton) About a million.
(Wolf) Are you going to have a million unit party at some point?
(Upton) We need to have some sort of party but the problem is we launched on the 29th of February, which denies us our first anniversary party, but we’ve got a great fourth anniversary party, so we should probably have some sort of millionth unit party. It is actually quite hard to measure once you get past a million units. You have to put your finger on what you are actually going to do with the millionth unit, because it’s going to be coming off several production lines, so it will be a slightly notional device out of a pool ten or twenty thousand that might be the millionth.
(Wolf) Who’s buying it? I know that you had this concept of putting it into kids’ homes and returning a little bit to this past. However I know that a lot of the interest is coming from the maker community, someone said the guys with white beards and fondly remembering the past. Who’s buying it today, are a lot of those going into the hands of children at this point?
(Upton) I would say probably 20 or 30 % of the total are going into the hands of children one way or another, either parents buying them or schools buying them, enthusiasts or even children buying them for themselves. I think that’s quite a good hit, and certainly we see vast numbers of children playing with these things and using them as platforms to learn programming on.
(Wolf) It’s a viable concept and returning to putting something into the hands of children to get below the OS level. However, …… wrote an article saying that today’s kids don’t care; they throw the iPod touch at their parents and tell them to fix it. How would you answer that: are kids wanting to get to the level that they can get at with Raspberry Pi or do they just want to code using some of the easy whizzywig tools at a higher level in the stack?
(Upton) There’s a spectrum, every time you step down a level of abstraction you will be leaving some people behind, and you are only going to get a tiny number of people programming an assembly language, you can program bare metal assembly language on this machine. I’ve seen people write operating systems from scratch on it. So the vast majority of people are going to be using the high level and we work very hard to run tools like Scratch and Logo, which are very high-level whizzy wig programming, advanced. But I think children are interested, and that’s the realization for us. It’s the risk, and the easy knock against us, that children don’t care about that, they just want to play Angry Birds. But the thing is, you could write Angry Birds in an afternoon. It wouldn’t look very pretty; it wouldn’t have the polish of the actual game. Your child’s not going to write Modern Warfare, they’re not going to write Call of Duty, but they might be able to write Angry Birds, and kids find making stuff exciting. Part of that is about providing a lot of whizzy wig tools, to give a lot of reach to the child, and to bring a lot of things within reach of the child.
(Wolf) You work with Mojang together with Minecraft, and that really was the masterstroke to get a broader spectrum of kids interested, since it seems to be a worldwide craze at this point.
(Upton) Yeah, I think that’s right. And this wasn’t a case of us lobbying Mojang; t was a case Mojang having a really interesting organization. I also think that they have a lot in common with us: they have had this level of success beyond what they could really have imagined when they were thinking about whether that allows them do interesting and new stuff. What they have done with the Pi edition of Minecraft is basically a port of the pocket edition onto the Pi. But what they’ve done is add a programming interface which effectively that allows you to connect to Minecraft from any programming language, Pearl or Python or Ruby, pretty much anything, connect to the site and then do things in the world, so it’s a great way for kids to get a little bit of interest in programming because they get instant feed back. Say you want to build a house: you have to build a wall by going brick, brick, brick, brick manually. But what you can do is connect to it from Python and you can go 4x equals nought 10, 4z equals nought10, put a brick and then you press enter and bang, there’s a wall. The nice thing there is that gives the child an instant introduction to programming, and the child learns about something in the Minecraft world. We’re just now waiting for the final release of that into the wild.
(Wolf) Great. What’s next for Raspberry Pi looking forward? I know you released Model A, what’s next looking over the next twelve months?
(Upton) Well we have Camera Board and that’s going to be a big deal for the maker community: it’s going to have a 5-mega pixel camera hopefully at around $25.00, that you can plug onto the Pi. We’ve got a lot of software engineering work going on, doing a lot of work to improve the performance of the platform, getting it to a point where it could be used more reasonably as a general productivity machine. It turns out there’s a vast amount of low hanging fruit, a lot of places where people have, either because they have been developing software for the PC or what, that they’ve been able to breathe out and there are loads of places where we can go and improve that, so we are spending quite a lot of money on that. For the foundation, a big focus on education, because we have been focusing on building computers, which wasn’t the foundation’s mission. So we need to tack back a little, start looking at education materials, outreach, and lobbying, because we need to start lobbying the government to get some the changes put in the government can do that, like teacher training. That’s the big deal for us over the next year or so we’ve been hiring to support that. We also have this Google project that we are running: Google made a donation a couple of weeks ago of a million dollars and that was towards deploying 50-70,000 Pis to targeted school children in the UK to try and bootstrap the revolution we’re looking for.
(Wolf) It seems like there has been more interest in Europe around Raspberry Pi from the government and private enterprise versus the US. Why is that?
(Upton) The differential I would address between North America and Europe has only really been about the stage of the publicity of that in these countries. In the UK we have massive support, we had a lots of support from the BBC fairly early on, so our earliest bits of press were with the BBC, so we went right in there with the mainstream media. Now in most other markets, it’s been the case that we had to go in via a specialist press and then raise awareness through the specialist press and then break into the mainstream press. We had a piece in the New York Times last week, and we’re starting to see a level of interest. Once the level of visibility develops in North America, I think we’re going to see the same thing that we see in Europe; corporate is going to find it interesting, the government is going to find it interesting. I think the struggle will be on a country by country basis the exact position of Pi within the education system is going to differ; in some countries it’s going to be more about after and decoupled from the formal education. In other places it’s going to be curriculum integrated, and it will be interesting to see which way the US goes, I would suspect the former, but who can tell.
(Wolf) It seems like a lot of the community that’s driven the Robot League interest, Lego League interest, but that’s just extra curricular, the after school programs. It seems like that could be a hot bed of activity with the same community there.
(Upton) Yeah, and the good thing there is that it allows you to bypass curriculum change, which is one of the hardest things to accomplish. By going around the side and appealing to the enthusiastic teachers who are committed, you can circumvent a lot of that. You still have to back fill and go and do the heavy lifting to make the curriculum work, but it means that you can get results two or three years ahead of what you get if you just plodded through the curriculum. Of course having stuff going on in the less formal environment gives you more leverage when you are talking about the curriculum that you can point to.
(Wolf) Where are you with Android? Is a kind of formalized Android release coming at some point?
(Upton) We don’t have the schedule for that. We’ve been working with the open source community to get an open Android. We could at any point push a closed Android onto this platform but we really would rather not, because you bifurcate your community. This business is for creating stuff. Android is an operating system for consuming stuff. We could write an operating system for consuming stuff, but it isn’t central to us. It is central to a certain group of our customer base and we want to support those people but our primary interest is going to be in the health of the primary platform, and the primarily platform is Linux.
(Wolf) What do you think the legacy of Raspberry Pi is? When you take the Raspberry Pis, do you hope a generation of computer centric programming types who drive innovation twenty years from now, that that started with Raspberry Pi?
(Upton) That’s my hope. If one person who looks back on the Raspberry Pi in the same way that I look back on the BBC micro, then I will be a happy guy.
(Wolf) Well Ebon Upton I appreciate your taking time here. I think you’re near the slopes so you probably want get some ski in at some point.
(Upton) I’ll certainly do that it looks like it’s going to be patter day today.
(Wolf) Well thanks for taking time.
(Upton) Thanks for your time. Cheers.