He started a computing REVOLUTION—then the shortage hit

this is the single hardest decision I've ever had to make in in my business career and certainly my time at Raspberry Pi what am I not sorry we didn't do I'm not sorry that we didn't float the price I think it's very unfair of you not to let me rewind history by three years because that's what you actually need to do in order to make a difference this month I interviewed Evan Upton at raspberry Pace Headquarters in Cambridge Evan is an engineer and the co-founder of Raspberry Pi now to be clear Raspberry Pi didn't pay for my trip the whole thing was financed by my patreon and GitHub supporters and by this video sponsor piesharp.us more on them later the first topic the pie shortage I think I think one of the challenges you have is that a lot of the um manufacturing capabilities will drop stream of us are running normally run very very close to capacity no one wants to have a factory in the horse have a semiconductor Fab say that's running at 90 capacity everyone wants a factory that's running at 99 capacity so you have a um I guess resilience resilience redundancy and efficiency they fight against each other what that means is if you get a little bit behind which is what happened as a result of some of the things that happened in 2020. if you get a little bit behind it can be almost impossible to catch up that of course has also been compounded the underlying shortage the Earth the underlying sort of true shortage I guess does as we saw with a number of other things like toilet paper in this country once there's a perception that their shortage people's behavior changes some people start to exhibit stock part particularly stockpiling Behavior people start to want to hold contingency stockhold components and so quite often the the actual underlying shortage gets magnified by some of the behaviors that people have in in response to uh in response to the shortage the shortages take a toll on the whole pie Community individual makers students and businesses that use pies but it's also been tough on the resellers resellers like pie shop.us who generously sponsored this video they actually have been getting more pies to sell lately and they've always had tons of stack of Pico kits they sell to home education and Industry and fun fact I bought my first Raspberry Pi 4 from them when I asked them about sponsoring this video they told me from their research seller's perspective the pie Supply is definitely getting better this year in fact they've been getting batches of Pi zero W pretty regularly and soon they couldn't promise an exact date but soon they'll be opening the floodgates for ordering as many Pi zeros as you need and that's great news definitely check them out I know I've bought a number of pies and accessories from them and their customer support has always been top-notch especially when I started asking about compute module availability way back in 2020 and I even got to talk with Evan about that people like me who did buy pies before the shortage started we talked about the idea of attrition in 2023 one of the things that has enabled people to sort of survive through the spirit of shortage is exactly that it's people it's that people came into this shortage situation holding lots and lots and lots of raspberry pies already so to some extent for a limited period of time people can survive a transient shortage by just as a a company might survive a Transit shortage by having some buffer of supplies and buffer of inbound componentry so hobbyists can survive for a period of time by using the raspberry Piers they already have it's interesting you go look on on Reddit though what you see is people are increasingly talking about attrition they're talking about I had I came into this with 15 Raspberry Pi's and I've now blown up five of my Raspberry Pi's doing projects now I'm down to 10 raspberry Piers if this carries all from then you can draw a line if this carries on for another year then I'm down I'm going to be down to zero or raspberry pies yeah so it's only ever uh that's only ever you know for companies or for individuals like that's only a temporary solution in in December last year you you said that it would be Quarter Two quarter three would be coming back quarter four you'd be able to go anywhere and buy one yes is that still something that that's about right so quarter one this year was our worst quarter in terms of production and shipment I recorded one partly because we pulled quite a lot of production into the Christmas period because we wanted people to have a reasonably good Christmas but quarter one this year was our low our lowest upper quarter I think since Q3 of 2015 right you've got to go back a long way we did about 750 20 hundred thousand units in q1 this year um and so you have to go back a long way before you find a quarter where we did that and it's incredible it's still incredible to me that I'm talking about that as being not many raspberry Piers Dr Evan of 2012. you sold quarter of a million half a million whatever yeah you're gonna be like wait and how many years yeah yeah that's right and one month I know there is a there is a kind of there there's a a misperception uh um imagine that because you can't freely buy a Raspberry Pi we must the only possible exploration of that must be we are making zero Raspberry Pi it's actually making a very large number of raspberry Piers you know sort of in a bad quarter you're not far off making ten thousand raspberry pies a day um but it's because of course you have you have backlog everywhere you have industrial customers in a backlog you have individuals who will um who are watching who are waiting in queues that approved resellers subscribe to our allocator um uh you know the demand our supply becomes available for you know until you get back to a really really healthy level of Supply which is kind of where we're getting this quarter even quite large amounts of production just get swallowed up by backlog as we get back to supply and demand hopefully equalizing like what do you think will be the volume at that point well it's a fascinating it's a fascinating question I don't know how big my business is right I haven't been I've been in constraint for two years now so we did six million units in 2019 seven million years in 2020 obviously 2020 was a strange year but it was non-constrained year for us so we were able to produce enough so we we entered that year we normally run about a half million unit backlog which is it's about months backlog it's fairly sort of that's that's not unhealthy um went into 2020 with a half million backlog came out of 2020 with a half a million backlog went into 2021 with a half million backlog came out of 2021 having sold another seven million units with the four and a half million in the backlog um and so you can you can't just add those numbers together so you can't oh yeah I mentioned um people's behavior changing when there's a perception that there's a shortage clearly that extra that four million Delta you don't just get to add seven million and four million and say we were in 11 million unit business if there hadn't been the shortage you wouldn't have got the four million units but it does feel like we were probably not a 7 million unit um uh business we actually have a very very healthy first quarter that year we saw it's all the sort of demand going up and then it was really that the story of that year was about not being able to scale Supply I up to meet increased demand and then sadly then last year the story was about actual Supply going down so last year we then did five million units and that's five million units even after kind of ducking for the line and we'd really duck for the line in December we still ended 5 million units um in q1 this year between 750 and 800 000 that's an annualized rate of barely three million units now the interesting thing about this year is we probably have capacity to do a 10 million unit yeah if we had to we probably won't have to but we could do a 10 million in a year but all but you know over if we did a 10 million unit year more than 9 million of that would be packed into the last nine months of the year so you you know that what that that year would look like three months 250 000 and then nine months of millions a month I think we are where we said we'd be in December which is fairly lousy first quarter um a second quarter which will be about two million units which is a pre-pandemic quarter which is great but obviously it's still catching up the backlog then a third and fourth quarter which are broadly speaking unconstrained yeah certainly by chip Supply I mean you actually will eventually Hit Factory manufacturing capacity limitations yeah I expect that we will have um satisfied or you know wound down all of our backlogs and got everything into General availability before we bump into a factory capacity limit and I have seen Pi 3 A Plus in stock pretty much in many places now pretty much continuously for several months now so the zero has been coming back actually I think there's a reasonable Channel Zero two might actually come back for good so zero is in a better availability position than it was um but zero two is I think gonna take um we've um remember zero two is powered by this this device called rp30 which is a broken processor dye and a micron uh memory packaged together by us into our own into our own custom sip um I actually got a lot of wafer Supply and Manufacturing test capacity for that now the interesting thing about zero two is because it's it's really a child of the shortage right it kind of came along just as the shortage really kicked in so it hasn't really accumulated industrial Demand yet it's too young and for most of its life it's been very very Supply constrained so it hasn't accumulated industrial demand so all of that Supply that's hundreds and hundreds of thousands of units of Supply they will all go into into hobbyist and education um so xero's got more Supply but it's also got more industrial demand zero two has smaller but still significant Supply but no industrial demand so I think those two will probably race each other three of us will stay in for for all time now zero zero two will race each other um and then we really start to see substantial sort of recovery in three B and 3B plus uh and then in pi four um as we go through into the into the towards the end of this quarter the pie blog post was pretty accurate just with a few slight tweaks to when certain pies are coming back in stock so I switched tracks and asked Evan about the level of anger that's present in the community right now a lot of people in my comment section say well I'm never going to use a Raspberry Pi again but I think for a lot of those people they would if they could get one yeah I think it's just they're there's anger do you know it's on the look it's completely understandable I I think I've talked about this a lot and this is this is the single hardest decision I've ever had to make in in my business career and certainly in my time at Raspberry Pi it is it's extremely hard to decide when you are a hobbyist as I am and you build built this thing for obvious and education to prioritize a different market right I think there's a it's really important to emphasize that when we talk about industry we're not talking about IBM we're not talking about multi-billion heavily Diversified multi-billion dollar International companies our median we have a few of those but um our median customer is about a 3 000 OEM customers about three thousand boards a year so these are often one two five ten employee mom and pop shops building some specific thing on Raspberry Pi so they can't get Raspberry Pi they have no real Alternatives um and they'll go to the wall they'll go bust right and so the real question when people are saying should you do this prioritization call is should I zero some of those customers or constrain them so heavily that they go out of business or take very very serious damage and it didn't feel like the moral thing to do um uh but just because I think we did the right thing um I think a lot of people recognize that we did the right thing or are prepared to give us the benefit of the doubt I hope they are um and and there will be some permanent disgruntlement but I I just I don't want I think it's important to address these misconceptions here the misconception that we're not making raspberry pies we are we're making lots and lots of raspberry Piers we're just making enough for Raspberry Pi's the misconception that we are selling to IBM that we're prioritizing like like business look because business doesn't exist right business is just people right uh and sometimes those businesses are quite small and quite fragile um and so we we made a we made a judgment call um I'm looking forward to not having to make that judgment call anymore if we could rewind history back a let's say a year or two years right what I think it's very unfair of you not to let me rewind history by three years because that's what you actually need to do in order to make a difference but yeah um uh two years if you like I say if you if you if you if you allow me to rewind three years ball 2835s would have been great this is the the cheapest of the chips that we use in Raspberry Pi products we could have afforded a stockpile and one or two million of those and obviously that would sustain zero in particular um uh all the way through and that would have been a nice thing if you know if you allow me two years I think what we would have done we would have got in front of we would have got this active management scheme in place a lot earlier uh we would have been much more proactive about getting in touch with our OEM customers and understanding their demand why is that important we actually tried to uh the that very anarchic put units in the market and the market will take care of it uh it's so core to who we are and so called how we think about ourselves from a business perspective we really didn't want to give it up but if you don't if you if you go down that route what you end up doing is you end up you end up empowering scalpers and see you know you end up with with boxes of units going out to people who've ordered them specifically to scalp them um and that's what that's kind of the point of the active management is that you call someone up to go on Zoom with someone you look them in the eye and say how many do you need and they tell you and you say yeah how many do you really need um I do think that that's been working because earlier in the shortage I was getting emails from people saying hey I have a box of cm4s yeah if you want to buy them yeah I haven't gotten one of those and that's horrible right because you know those are units that are you know maybe they're still in someone's shed while they wait for the right price I hope they still I kind of slightly now because we're so close to the end of it I kind of hope they are because I hope the person then gets rinsed when uh when they try to analyze it the toilet what am I supposed to do with this now you know um so you know yeah when they turned off when Amazon turned off um uh Amazon's Marketplace and eBay turned off hand sanitizer sale so these guys like but I have an entire house full of hand sanitizer well you know tough you shouldn't have done that um so yeah but yeah maybe these are still in someone's shared waiting for the optimal moment to sell maybe they got bought by some company that was desperate when you know if they if we could have held them to ourselves we could have had a relationship with that customer with that company uh and we could have because we haven't put prices up around but we put up um a pi42 gig went back to 45 uh all cm4s went up by five dollars um and then we recently adjusted zero and zero W but that's the entirety of our pricing action even inflation you guys are still keeping down yeah we're below inflation on almost everything right you know um and you know we're really married to those old to those old price points um and and so yeah what we haven't done is we haven't gone for the ultra capitalist kind of float the price because there is a you know what what am I what am I not sorry we didn't do I'm not sorry that we didn't float the price and then let capitalism take care of it [Music] um yeah oh my God you know I mean you know you could have you know the amount of money that we could have made if we'd float over the last couple of years um and you know if we if we'd if we'd floated the price of the products but I think that would have been a horrible betrayal of trust I know how I feel when people do that to me um so I'm glad I'm glad we didn't do that I'm glad we went for the it's been a hugely labor intensive when you look at the commercial team here and the amount of effort that they put into you know this is not an organization that is was designed to maintain thousands of individual relationships with customers that's just not what we built but fortunately we had people in the commercial team here who have the aptitude call it anticipal someone got who says management yeah yeah yeah yeah Tom mcclusion says I want a thousand of something usually you would expect a good salesperson to say uh once you've got maybe like 2000 these days our sales people say would you like 500 you know do you need did you need a thousand when you say a thousand are you going to build a thousand units tomorrow or can I give you 100 units now in 100 units next month right just as the communications uh function is had to Bear some of the brunt of the of the kind of unhappiness about about these decisions the uh the commercial team has had to Bear the brunt of the the kind of the labor of maintaining these these active relationships with oems switching cracks completely from the shortage to something that's in great Surplus the rp2040 and the Pico yeah what like that was some dumb luck right yeah when I made my video about it I I saw like oh it looks like an Arduino and that's you know maybe a little bit different but what has surprised you the most about that product um I well as you say it's been lovely to have a product which is in um great Surplus it's it's been good to have let's talk and talk about some dumb luck right you know if you had to pick a month for this thing yeah what yeah what did what are my regrets um RP 2040 came into availability volume availability in the first quarter of 2012 so about 12 months after the um Pico launched if I could have had six months earlier than that I think we could have made much more contribution to helping people through the microcontroller bits of the semiconductor shortage but I mean I just I I guess I've been surprised by how quickly and some of this is a function of the shortage environment um I've been surprised by how quickly people have been prepared to adopt and an entirely new architecture right you know it's an arm chips has a lot in common with other modern microcontrollers um but a lot of the architectural decisions we've made which I think are good decisions um make it unusual uh as a platform I've been super impressed by how everyone not just the kind of close end guys we had a few people who we provided with um Early Access so we had Adafruit Arduino which is lovely you know we love Arduino awfully much very inspired by Arduino so our Adafruit Arduino primaroni and Sparks and all had early access to the silicon and they did predictably awesome stuff with it out of the gates yeah yeah those are I was super I was I was I was super happy to have that there's a fun thing with Arduino where there is other misconceptions there's a misconception that Arduino and Raspberry Pi are competitors uh and it's hardly ever been actually because we're in quite separate spaces even now we're making microcontroller-based product um we're actually in very different spaces in terms of of of of um if you have a particular problem it is very seldom even now Pico and Nano 2040 exist um it is very seldom the case that uh there's genuinely any confusion as to which of the products you should use to solve your problem and so yeah we've been friends with them for a long time we're very inspired by them they will they along with olpc were probably the two kind of Hardware companies Hardware startups that kind of inspired us at Raspberry Pi and so it was really wonderful to be able to see an Arduino product an Arduino platform which has been very successful um are built around our silicon the nice thing about the rp2040 and the Pico is they've been available and since Raspberry Pi designed the chips themselves could they someday move to risk 5 and save on arm licensing fees I really wanted to get another perspective too and I found out that Ian cutris from tech tech potato lives in London in fact I'm having a Cuppa in one of his fancy mugs right now he's an expert covering the chip industry so I asked him what he thought about the prospect of a pie on risk 5. the whole point about risk five is that it's this open source ecosystem where anybody can add anything now unfortunately if anybody can have anything it means everybody doesn't support everybody else yeah um so there has to be that next level of standardization arm already has that risk slide is getting there just not yet yeah um so I wouldn't fundamentally ever say Roger Pi would pivot to risk five if if they were to go down that route it would be the add-on so what better follow-up to that than to ask Evan himself what are your thoughts on whether that is something that could be in Raspberry Pi's future have you guys been involved in any kind of risk five design well we're uh so we're a silver member of risk five International um so we're a silver member of the the rissified foundation it's interesting there are so I'll split the world in two there's what we call the a class here we just use I mean almost so dominant now you use arms words for everything so you have the A-Class base in the m-class space you know the big cause in the little balls here um in the A-Class space there's there is a I get I tend to get I tend to get flamed a little bit for saying this but uh because people people will say oh but on GitHub you can find this excellent core that is much more performant than I think I'll make but um there really is a shortage of good licensable high performance chords really high because remember we're shipping an a72 in Raspberry Pi 4. so if risk five was going to go into a future Raspberry Pi you need something which is much much more capable than a72 and they really aren't many cores in that space there are a lot of a55 plus cores the in order cause so there's a shortage of licensable cores there's a still a lack of maturity in the software Stacks um in the in the um uh the particularly bits of the Linux use land are not well optimized at the moment for uh for risk five architectures not to say that these challenges can't be overcome because there was a time when arm had these charges you know you compare arms 20 years ago yeah yeah 20 years ago I mean it's still the case that if you go through pixman say you know some some some library of ffmp pics man whatever um and you look for fast paths the place is crawling with Intel fast Parts yeah everything from mnx to SSE and onward there are fast paths for every generation of of Intel multimedia acceleration instruction um and on the arm side you have a handful of Cortex A8 stuff that Nokia did 15 years ago and then you have a trail of arm 11 uh cortex A7 codex a53 and cortex a72 optimization see what you can see what those all have in common um that some other company contributed over the years so you know even the arm world is pretty immature compared to the Intel world the the risk 5 world is immature then compared to the armor that could be overcome an arm overcame it to us if it not to agree not completely but to a sufficient degree uh I'm sure risk five can do that but it's going to take years and there's a chicken and egg uh there's a chicken and egg problem you know arm are very very entrenched in that space microcontrol is more interesting yeah right because the software Stacks are shallower the course are simpler the software Stacks are shallower there are a lot of places you can go and license an m-class um risk risk five core and so I do think and you are seeing that now right particularly the ultra low end um I do think there are opportunities to uh for people to go build risk five um microcontrollers uh would we do it um I don't know I'm in the arm the armed value proposition is really strong right it's a really strong community and it's not expensive to play in it I've been working with an ampere ultramax system yeah I've seen I've seen your tweets it's been very fun having that much power in that small in the space in the arm world and then you know the M1 M2 chips yeah we were really pleased to have launched Pi 400 the week before the M1 Max because it meant that they are as a child I'd lost it over these Acorn Archimedes machines which had arm twos in them um and then and then we they were then absent for 20 years from the desktop space and we got back in a week we were the first back in the door by about a week before um the other fruit company came out that was nice my kids are learning Computing on a pi 400 yeah it's a lovely platform right and one of the cool things about the one of the cool things that's happened over the last few weeks I think is that 400 is coming back it's not extensively tracked but it is I went to Micro Center and they had Pi 400 yeah they had I think three or four in this desk yeah and they said that they had gotten a few pie fours a couple weeks before yeah right now I'm not buying any because I want other people thank you if you're uncustomed of this you know it's like everything is is we're an upside down world at the moment um uh but but yeah it's um so I think I would some kind of Say Never Say Never I think microcontroller is more plausible than a Class A Class will may become plausible in a few years um but M classes m-classes definitely is definitely feasible and I I definitely wouldn't wouldn't commit to not do it I actually posted a lot more about the idea of risk five and the pi on my website so if you want to go deeper check out the link below Raspberry Pi might not be fully committed to risk 5 but they are fully committed with Sony now judging by the fact that Sony bought a minority stake in Raspberry Pi this year I asked Evan about it we've been working with Sony for 11 years in contract manufacturing um so we are the biggest the oldest and still by Revenue vastly the biggest um relationship we have with Sonia's contract manufacturing almost every uh Raspberry Pi products I mean every core Raspberry Pi product is now manufactured for us by Sony uh most of it in uh South Wales in pencoed um so that's the oldest relationship we have an image sensor relationship with dates back to 2016 so we have what is whatever that is now a seven year relationship the V2 camera those are the V2 cameras the first camera was an omnivision camera that sensor got elled and then we did the 219 camera it's our only camera product for a long time and then we did the HQ camera with 477 in it um and then the V3 camera was 708 and then the global shutter camera with 296. because my brain is just full of numbers are you planning are you planning on becoming a camera company at some point well that's it I mean the interesting thing is you know those are they're great products right and they're I mean they're great products um they're they're yeah they uh yeah commercially you know we make money from them um they also make the platform the the overall kind of Raspberry Pi platform the Raspberry Pi ecosystem stick here because it's where these nice cameras are you actually see people figuring out how to plug our cameras into other people's into other people's spcs which is kind of fun yeah and I never feel I never feel bad about that you know I'm always happy to happy to take the margin from that and get Raspberry Pi logos in all places um but yeah they are great and I mean Sony Sonya so Sony have been a great partner and then they've they've got some really nice capabilities in AI we've never we've never integrated an AI into the core product AI acceleration the core product I don't think we ever will yeah um and the reason we won't is because it will only ever be of interest or minority but potentially a sizable minority Bill a minority of our customers it's quite an expensive feature to put on the board and if you think you're taxing every customer to pay for a thing which is only of interest to 10 35 price point yeah I can't just throw it on yeah yeah that's it I mean you told see I mean the sort of the the most incredible example of this it Remains the lack of an ADC on a Raspberry Pi you know an ADC is a 20 cent i2c component um but we've know it's never made it onto the pie rise it never made it onto the pie because if it's only of use to five percent of people then you kind of count a 20 a 20 cent component we also divide through by the fraction so you imagine a a 20 cent competency used by five percent of people is a four dollar component in our in our reasoning the only what's the only capability that a modern Raspberry Pi house or the Raspberry Pi one didn't have wireless networking right that's the only uh uh qualitative difference there's lots of quantitative differences and one more memory faster processors to HDMI ports lots of lots of those things the only actual quality that have changed to the platform that ever made it through that filter is Wireless networking so I don't think AI is going to go in the core platform but there are entities using some of the products Sony has to add AI as an accessory product to the pie this year everyone's doing AI on text but um historically what has everyone been doing AI on people have been doing AI on images with what they've been capturing the images with a camera so it probably makes more sense to have an integrated an offering which is an integrated camera with AI acceleration and then a non-ar accelerated baseboard than it does to to if you want a two box right now you can do a three box solution you can buy an HQ cam a Raspberry Pi and a coral accelerators you've got a three box solution if you want to go to a two box solution there are two partitions one where you put the acceleration on the core product can't do that the other one where you put the acceleration on the image capture device could do that and I think that's probably what this opens the dot so AI camera coming when oh yeah yeah it's not a product announcement but but it's suddenly you know you look at that lineup of you look at that you look at their product lineup and I think there are opportunities to build really really really interesting products there so I also asked especially if half of all Raspberry Pi's end up in Industry do you prioritize things like Edge compute when you're building the next pie are there conflicting priorities like commercial versus educational use the good thing about the product design is we've never had to prioritize in that way you know people say well what decisions did you make for yeah we have compute modules so we have an actual platform you've got Pi 400 and compute module so you've got the the parent product the SPC and then you've got something which is obviously derived for industry and something which is obviously derived for Consumer but hey we use a lot of the the chord most of the money gets spent on the core platform always trying to educate people that's when you're doing a development that's where the money goes but the core platform we've never made a decision I don't think to Target one or the other again the core platform is still the bulk of our business so yeah you know seven million a year we're probably doing five million units of the of the SBC and they're selling into both markets there are no decisions that you make to make it specifically useful for industry or or specifically useful for a for a hobbyist and I think make it better I think the the interesting thing is I've from a lot of the business users of it including my dad who doesn't he uses them for remote access at tower sites the whole reason they like it is because it's so easy for a hobbyist or a maker that it's also easy for you to pick it up and use it in other projects it's actually the same they're the same capabilities they're the same um uh you know they're the same the same platform capabilities that you sold to both groups um and often you know both groups the both groups are actually the same people you know the hobbyists you know why is Raspberry Pi so popular in industry is popular in Industry because it's popular with hobbyists who happen to be design engineers and they take it with them into into into their professionals or people that are trying to build a petabyte storage solution yeah that's it that exactly exactly those exactly those it's actually kind of important there is that there is an angle there which is it's actually important you know these are you know we're talking about businesses even even when we're talking about somewhat larger businesses you are you are talking about individuals if a larger business is designed to Raspberry Pi into a platform someone made that recommendation right somebody in the organization stuck their neck out and said to their boss you should use this platform there's nothing will go wrong if you use this platform those are real people who would have it's another classroom people not just your mom and pop business owners that's another class of person who would have been very severely impacted if we had then either by pushing through price changes or by zeroing um oems um those are it's another class of person who would have been personally impacted by that decision it's not to say it wasn't a decision it could have been made um but it's not a it's not a Costless decision and it's not it's not a decision that only impacts uh corporations uh switching tracks a little bit I asked on my Discord if anybody had any questions and the best question was why raspberry oh uh raspberry um raspberry is fruit named computer companies there have been several uh large computer companies named after fruit uh we had an apricot in the UK we had a tangerine actually here in Cambridge Acorn is I think a fruit um uh and so yeah there's lots of there's lots of fruit and computer companies I think I might have missed another one out somewhere um so there is that a raspberry is also the rudest is also the rudest fruit um because I don't know in American English if it's if it's like no it's a blank of raspberry um so and we and yeah we're talking you're talking to talking to Children you know they um Roald Dahl had this thing where he said that they you know uh the the the you know the best thing for a kid the funniest thing for a kid is when an adult thoughts right um and the funniest thing that could ever ever ever happen would be if the queen were to fart um and so yeah there's a there's a certain kind of purol um uh uh love of of raspberry for that reason and then Pi of course is python uh pies Python and also a feeling that the the letter pie would make a nice would make a nice logo and we'd end up with a picture around with a picture of a raspberry and you have an annual day of Celebration we do have an annual day of Celebration and only uh I mean obviously I celebrate it only because I'm an engineer not a scientist I celebrated on the um uh the 22nd of uh the 22nd of July it could be worse I could celebrate it on the 3rd of January I suppose before I leave and conclude I have to say can you commit to supporting PCI Express better on the next generation of Raspberry Pi whenever that comes out yeah I think we can um uh in the 64-bit thing is not ideal I it really upsets me that we can't drive the the Google TPU um using the code I mean it upsets me that others it doesn't upset you that I can't drive a 40 90. um it's I it upsets me a little tiny bit um uh yeah I I yeah when you say better um I'm gonna be fascinated because because you know that really is kind of like pushing right at the limits you have to remember the piece of Express controller that's in 2711 was designed to drive Wi-Fi chips right the specific companies Wi-Fi chips a a large manufacturer of Wi-Fi chips um yeah that's it I worked with it worked with vl805 which was a good which was you know a bit of good luck um uh but you know it's not by any means a PC PCI Express um controller um and uh yeah we've been happy with a range of devices that it will drive um yeah I think we'll try and do when whenever we do and interesting thing about Raspberry Pi 5 would be knocked back you know this is a this is a product where um you know you know what our product release cycles look like um before the pandemic before the pandemic you know if you it does still feel on some level like it's April 2020 and if it's April 2020 then Raspberry Pi 4 must be 10 months old right um so you know we'll be knocked back a little bit um but yeah I think we we're now getting to a point where we can actually think about the next thing um a little bit more what do you want the legacy of Raspberry Pi to be oh I went on the Legacy to be oh that's really really simple um I one of the nice things for me about Raspberry Pi is um uh for me uh from just a personal perspective is that I got to meet most of my heroes so I got to meet the people who built the BBC micro which is the machine I had when I was a child um and um I got to say thanks um uh and if I had the good luck to work with Sophie Wilson who's who was both the person who wrote BBC basic and also the person who designed the instruction set for the the 32-bit arm instruction set oh I got to work with her for many years um but I've met almost all the acorn people um and so I'd like my legacy to be I'd like that to be one person who looked back on we sold 50 million Raspberry Pi's now right um and millions although you know probably industrial consumption even you know in an unconstrained here industrial consumption would be well over half of Raspberry Pi is that still millions and millions and millions of units going into the hobbyist in education community I would like it if there was one person out all those people who is has the affection for the Raspberry Pi that I have for the BBC micro because I owe everything uh including having met Liz who I co-founded this thing with um uh I owe all of that to the little beige box in the corner of my bedroom that goes a bit when you turn it on and gives you a basic prompt well the raspberry pay has certainly made an impact the last thing we discussed was what's coming in the next 10 years of Raspberry Pi the franchise warm if you thoughts think of of making electronic products as being the first the thing we installed accounts thing we did first is the thing that counts the majority of our business um uh if you think of that as being a franchise one and you think of doing semiconductors rp2040 as being franchise too even in franchise one there's a lot of cool stuff to do even before you start thinking about the new things you unlock if you allow people to buy if you put people in a position to buy um uh the the chip the powers of Raspberry Pi Beacon all right well thank you very much and speaking of the Legacy I do admire your work but I am not somebody who grew up on well that's it I already had my passion what I like is I mean obviously your somebody who's always had a passion for it I think that I found um I think that I found really heartening actually is somewhere like the UK where you where there was a big ground swallow of excitement about Computing in the 1980s um when you meet somebody who lifted away from it and you know we do have people you know certainly in the early days um people who would say man I used to love this when I was a kid and I just stopped doing it and now I'm doing it again and it makes me happy yeah that's it that's it that's a that's a that is a common refrain it's a fun it's a fun thing right so and you know um you know who can tell where we'll be in a year's time yeah well thank you very much well no well thank you though thank you for I mean like you say you've come you probably hold the record for having come the furthest to see pyth hours it is cubicle and but uh it's cubicle happens it looks like it's expanding this room that we're in it apparently it keeps being pushed back further yeah this room is this room is probably you may be one of the last people to experience meeting room three and it's and it's uh hotter sibling meeting room four um uh before the thing gets turned into more cubicle land so anyway well thank you again yeah thanks thank you and I'd like to thank Raspberry Pi for allowing me to visit all of you for watching pie shop.us for the sponsorship and a huge huge thanks to everyone who supported this channel on patreon GitHub sponsors or YouTube memberships I couldn't have done this without you nor could I have visited the Sony Factory in Wales where they make raspberry pies subscribe because you definitely don't want to miss that video until next time I'm Jeff geerling

Channel: Jeff Geerling

Views: 222,071

Keywords: eben upton, raspberry pi, interview, jeff geerling, uk, one on one, shortage, pi, component, supplier, sony, factory, british, computing, history, legacy, makers, commercial, business, pishop, reseller, microcontroller, rp2040, bcm2711, chip, chips, silicon, ian cutress, techtechpotato, towers, headquarters, blowing, dahl, bfg, farting, queen, aitrios, semiconductor, manufacturing, pi 400, arm, risc-v, semi, apple, acorn, cambridge, tangerine, bbc, micro, camera, community, response

Id: -_aL9V0JsQQ

Length: 37min 44sec (2264 seconds)

Published: Thu May 18 2023