 I do electronics. I do much less electronics than I'd like to because I can't get parts. And I suspect you're all suffering from the same thing. So the thing is, yeah, since late 2020, an increasing number of parts have been impossible to buy. The lead times, that is the time between when you order something and when you actually get it, has gone up from what used to be about one to two months to more than a year, sometimes over 80 weeks. And those lead times keep shifting backwards as well. So even if you order something now to be delivered a year from now, it may actually be delayed. And this is new, it's scary, it's never happened before in the entire industry. So let's try and understand it and see how it happened. So I'm going to give you a bit of background on how chips are manufactured and how the electronics market works. And then we're going to go over what actually happened that broke things and why they are so difficult to fix. So the electronics ecosystem, you have chip designers, you have foundries, you have chip vendors, distributors, product manufacturers, and users. So chip designers design chips, they send their designs to foundries, then you have chip vendors who either own or rent foundry capacity, and then you have them selling chips to either distributors or to manufacturers directly. And manufacturers manufacture products and then they sell them to users. That's how it normally works. Foundries, this may be a new thing to some of you. So foundries are a specialized kind of factory that basically is designed around edging silicone wafers. So they'll put some layers of a circuit into silicone wafers and add a lot of metal layers on top of that silicone and that's then ready chip. And each of those production lines is built around a particular scale. So you can have chips with smaller things on them, smaller transistors on them, and you can have bigger ones. And they can't all be used for all purposes. Now an important thing to understand here is that this process takes a really long time. It can take months between when you actually put in a chip order to when it actually gets ready to package. So you need to plan these things in advance and you need to be looking nine to 11, sometimes even longer months ahead. Now about these process nodes. So these are the scales of things on the chip. So smaller things, they can switch faster and they use less power. And bigger things can have no more voltage, more current and more interference, electromagnetic radiation, voltage variations, that sort of things. And bigger things are easier to make. So they have better yield. So less of it goes to waste when you manufacture it. The smallest nodes are what processors and GPUs use. That's the ones that you see in marketing. You have five nanometers, seven nanometers, this sort of thing, which chip manufacturers really like to make advertising with. But microcontrollers, for example, use bigger nodes. And most of the ICs in the world actually use even bigger nodes because it's much, much cheaper and it's much easier to manufacture. So you pretty much everything around you uses microchips that are made on fairly large process nodes. And the problem is nobody's actually making large process node equipment anymore. So what happens is you build a production line on the latest greatest thing, which is the smallest possible node, and then you need space to put that in because those things are gigantic. So these are entire complexes of buildings. So they're like, okay, we have this new and greatest thing. We need to get rid of everything that we already have. So let's just sell it down to somebody else. And then they got bought by a smaller foundry who doesn't have that much money and they're less famous and they have less market power. And they get to manufacture the larger process nodes and they get much less profit out of it. So they have lower margins. And nobody manufactures large node equipment anymore because it's much more profitable to be going with the latest thing. So all this larger process node equipment just gets passed down and there's no new market for it. So this has been working for several decades fairly well. There's never been a ship shortage. So yeah, let's look at what happened to break it. So in early 2020, a bunch of car executives have a conference call and they go, look, there's gonna be a recession. Everything's shutting down. People are going to buy less everything. They're gonna buy less cars. We should make less cars. And yeah, car manufacturers are a bit of a special thing in that their contracts with their suppliers are extremely flexible for the car manufacturer. So they don't commit to buying anything. They just pay for what they buy. So they co-op all their suppliers and you go, hey, you know what? We're actually not going to be buying anything this year. So you can produce it if you want. We're not going to pay for it. You do whatever you want. So all their suppliers take a massive sigh and go like, okay, this is gonna cost a little money. So what they do is they co-op all their foundries and go, actually, I know we ordered all these wafers for this year, but we're not going to need them. Sell them to someone else. And I know we have to pay a penalty for this. It's how it is. And they reassign all their foundry slots to everything that's not the stuff used in cars. And everything that's not the stuff used in cars is still fairly useful stuff. You have power management ICs. You have logic ICs. You have all sorts of things. And yeah, the entertainment industry was very happy to take those. They can make more TVs. They can make more playstations. That's great. Right. So the foundries get a bit upset at this because they have to cancel things mid-process or they have to throw away some stock that they've been working with because it's now too expensive to actually turn it into a finished product that nobody will buy. So a bit of manufacturing capacity is lost here, but this happens every once in a while. So that's fine. So a bit of background here to the car industry. So at some point, the car industry developed this management philosophy which has turned into a sort of religion to not have any inventory ever. So this was a thing designed by Toyota originally. And they came up with this because they had all these factories around them in the region near them which were delivering parts to them. And they figured out if we actually don't get these parts delivered until we actually need them, it's much more efficient because we don't need to store them in our warehouses. So if they're next door, we can keep them in their warehouse, bring them over when we need them. And that's fine. And this has kind of grown a bit into a kind of like point of truth for the car industry that like never keep any inventory ever because this worked so well for Toyota and Toyota was really killing it in the 80s. So like we all have to do that. And they all do that. And they can do this in a situation where basically you have a global supply chain and things get chipped across the planet because they can then just push that inventory risk onto their vendors and go like, okay, if you want us to buy 10 million parts from you, you're going to have to deliver them within this 15 minute window and we're going to tell you a week ahead how many we need. And you take that or you leave it. So yeah, this is really risky for the part vendors because they may be stuck with a warehouse full of stuff they can't sell. But yeah, that's how it is. So they figure, okay, like demand for cars doesn't really change that much year over year, right? So like, yeah, we can take the risk and we'll figure it out. So what happens in 2020? Yeah, people don't like public transport and they work remotely, so they move out of cities. So they're buying more cars than ever before. So all the car executives have a panic conference call and go, hey, all these parts we canceled, we want them back and more and more. And yeah, all their vendors go, yeah, we can't make those. Fuck off. We're going to go out of business if we actually need that. So the car companies go, okay, we have no inventory. Our production lines are stopped. We're screwed. We can't make anything. So yeah, I know we have this really asshole contract, but we're going to actually pay you anything you want. So all the component vendors go, okay, great, we can do that. So the market price settles at about eight times the original prices and they just cancel everything else and just start producing correlated parts. Which is great for them, because then they can slightly recover their losses from the extremely low margins that they've been having to deal with. And all those non-canceled top slots get reassigned to those. So everything else stops being manufactured. And because you're changing mid-process and the process that takes months, you end up with a lot of scrap. So all the stuff that was being manufactured at that point gets thrown away. It's like, okay, start over, we're making this now. Right, great, not that great. Yeah, at the same time, we have a few disasters. So in Texas, Texas is a really popular place for foundries to be, because they don't give a shit about pollution and they have occasionally very cheap power and they're very, very industry friendly. So a bunch of foundries are in there. Yeah, Texas has terrible grid management. So early 2021, Texas froze over. A bunch of water pipes in LXPs and Samsung's foundries in the walls of their clean rooms broke because they froze over. They sprayed a bunch of dirty water into the clean rooms. All the production was lost, factory was lost. Like, yeah, they lost months and months and months of production. And in Japan, there's renaissance. So NXP is the number one biggest automotive IC supplier. Samsung is number three, renaissance is number two. So renaissance, their main factory in Japan, they had a fire, they got smoke in their clean room, contaminated all their wafers, they had to throw all their wafers away. Yeah, AKM, manufacturer of audio chips used in all sorts of audio equipment. They had a fire in another factory in Japan, they lost their factory. So all these things are happening at the same time. So yeah, early 2021, some effects of this start becoming visible. So some parts are starting to run low on stock. Thankfully, we have distributors and distributors are great because they will buy up parts in bulk and then resell them in smaller quantities and they can kind of level out these effects of the market kind of moving around. So they still have some stock and this is great because then you can still buy it from them and you can make things. But some parts are starting to run low on stock and they're running lower than they've ever been before and lead times are going up a bit. And yeah, some parts, the ones that are used in cars have massively gone up in price and this is really weird. This has never happened before in the industry. So everybody gets a bit panicky and now everybody who's making anything with electronics has a problem. And the problem is electronics devices have many, many parts in them. And if one of them is missing, you cannot manufacture the entire thing. So normally this would mean like, okay, we fucked up with our ordering, we need to wait a month or two before we can produce more but we have some stock, we're fine. But yeah, now the lead times are jumping up to 16, 18, 20, 13 weeks which has never happened before. So yeah, you panic and you buy everything in advance because if you don't have everything, you can't produce anything. So you buy as much as you can as soon as it's available because you have no idea if you'll ever be able to get it again. And of course, if you're a small electronics manufacturer, you can't afford to do that. But hey, it's low interest rate season so borrow a shit ton of money, buy everything because otherwise you're out of business because you have no product to sell. And a bunch of products get actually redesigned around parts that are more available which makes those parts less available and there's this wonderful cascading effect and at the end there's just no stock of anything anyway. So every microcontroller is gone, all the memory chips are gone, a bunch of logic chips are gone, voltage regulators, like even the things that like you never imagined would ever be a problem, like a 7805 voltage regulator, it's gone, nobody sells it anymore. So all these things just disappear. Right, and the problem when the market just goes to zero is that the price and availability signaling system that normally runs the market just fails. So everybody's hoarding because they need to hoard because otherwise they can't produce. However, nobody's actually making more stuff. Actually they're making less stuff because they can't really afford to take financial risks because they've just put all their financial risk budget into sourcing components. So all the chip manufacturers know this and they know at some point this hoarding will eventually end because everybody has all their parts and they don't need to buy anymore for the next five years. So if you're trying to set up a new assembly line which will take you 11, 12 months to set up and yeah, by that time, nobody's buying anything. You've just paid for your assembly line and you've just entirely lost all your margins on everything you made and that assembly line will just be idle. So chip manufacturers know if they mistimed this, they will be ruined and they don't know what the timing is going to look like because all their market signaling has just gone away. So everyone's doing what seems best for them because they have no information about the future and they're all operating in the dark. So yeah, this didn't work out great for the car manufacturers either. So they had to all stop production this summer with the exception of Toyota who apparently is not that religious about not keeping stock because it doesn't actually make sense. Like you can get all the ICs that a car factory needs for two years worth of production can fit on this stage. It's not much. It's not like you're storing stacks of car body parts but nope, nope, we can't keep stock just in time, that's the way it goes. Yeah, so car manufacturers panic and they go like, okay, let's force our governments to force the chip manufacturers to only sell to us which didn't work because like there was nothing to sell. So that just screwed them over and basically every car manufacturer in Europe was having their production lines completely stopped this summer. Yeah, last summer. Right, so the political response to this was we must construct additional foundries. We cannot be dependent on like foundry capacity being this low. So yeah, who's making foundries? Intel's making foundries, let's give them all the money. Turns out nobody's actually making foundries for the things that are in shortage. Nobody's even making foundry equipment for the things that are in shortage because this stuff was last manufactured in 2004 and like nobody's making it anymore. Nobody's making 200 nanometer exposure machines anymore. It's just not profitable. You make way more money going to 753 nanometer. So nothing actually changes except Intel got a shit ton of money for nothing. So Intel actually went into on the new cycle going, oh yeah, car manufacturers redesign all your stuff to work on 14 nanometers and we can build as much as you want because we've got those. And like, yeah, okay, like if you can get car batteries down to 3.3 volts, that could work. But yeah, nope. So yeah, how to fix this? The core problem is you have no visibility into the future. You have no idea what's going to happen. Like availability of something, it's just going to change at some point in the indefinite future and you have no idea when and if you mistime it, you're screwed. So there are some solutions to this. So for example, Microchip sent a letter to all their customers, I got it as well. And they went, hey, we know this is shit. It's shit for us too. How about you tell us how much you're going to need a year from now of each part you're going to need and we'll provide them 11 months from now. And if you change your mind, you don't get your money back. So yeah, this is great. You commit to something. You know what's going to happen in the future as long as they actually deliver. So a customer of mine had this with another vendor who I will not name, but it's a big tree letter electronics manufacturer starting with N. So they had a commitment to get things at by a particular date and then that vendor decided to just give it to somebody who gave them more money. So yeah, it's like that's not how you manufacture trust. It doesn't work. So the problem is like the trust in the market is something that is really, really difficult to recover. The manufacturing itself has by now recovered. Like we these short shocks to the manufacturing process which happened from changing things around rapidly. That's already been recovered from, but the confidence in the market is gone. And it will take a really, really long time to get that trust back again. And until we do, we're all operating in the dark. So there are some medium term solutions. So you can redesign things around availability. I have been doing that with customers and it's been an absolute pain because you, yeah, you design a thing because a part is available and then you do one production batch and then it's not available anymore and you have to design it again. So in one case, we had one product which was really designed four times, different part every time. And yeah, each one got one production run with the maximum number of parts available at the time which was not very much. And yeah, there are some vendors that are less affected. So this has been an issue for hacker camps as well this year, especially the ones that have badges. Yeah, so they've been forced to use vendors that have better availability. So for example, the badge for this camp has chip by expressive, which are much less affected because no car vendors use them. So that's great. They basically were not screwed by this at all. So they could just continue production as normal. A lot of people are actually buying parts before they design. And this is what I'd recommend for hobbyists as well, just buy the things first and then design your thing around them. It does happen on a larger scale as well. So I had one customer call me and go like, oh, we've bought this many thousands of this part. And we hope you can redesign our product to work with it. So yeah, that does happen. It's just the sort of thing which a year earlier you would never have imagined hearing. Yeah, and there's also some vendors that are starting to manufacture function compatible parts to sort of substitute for the gaps in the market for vendors that have been severely screwed. So you have, ST was very badly affected by this. So STM32 parts, for example, are completely gone. And there's several vendors. There's WCH and there's GigaDevice who are making STM32 code compatible and pin compatible parts. And that has saved a lot of people. Yeah, but those are very temporary solutions. So medium term solutions, we need to reduce the fragility of the system by not screwing the vendors on contract terms. So if we can optimize for trust, then this will stabilize the markets and everyone will make more money even if they're not optimizing for making more money. And we need to not support bullies. So the reason that the car companies can get away with screwing their vendors so badly is because they have the financial and political power to do so. So they can go, okay, we'll pull this contract and your company goes out of business. So you do exactly what we say and they can get away with this. And I don't think that's right. Right, yeah, I do have five minutes left. And I think I'll stop here and take some questions. Hey, if there are any questions, please line up at the microphones in the middle. In the meantime, does the internet have any questions? Not yet? Okay, then microphone in the back please. Hello, you said in your talk that there was nobody to be made for 200 nanometers fabs. I'm not mistaken. That was the assumption at the time, yes. I've read in some press, I think it was Bloomberg or something not specialized, I mean, that there was, very sorry, a lot of companies that are currently building fabs in India, especially for these common parts. Can you confirm that rumor that? Yeah, so lots of people are building fabs for those processes, but they're built with all the equipment. So it's very difficult at the moment to actually get new equipment for anything but the latest knowns. And it's actually very difficult to get new equipment for those as well, because the capacity for the equipment manufacturers is completely gone. Okay, then microphone in the front please. Hello, so we explained to you how this problem occurred for most of the silicon components, but we've even found that for one of our projects, essentially we are running out of basic things like plastic connector housings, which are spiked in price like one euro per basic connector. Is that still a part of the same panic box? This is the same thing. So like if you're missing one item on your bomb on your bill of materials, you cannot produce your whole product. So you buy absolutely everything you need. Like some people are having trouble getting particular lengths of screws. It is really that bad. Okay, microphone in the back please. Yes, so we have this list of medium term solutions on the projector now, but won't these raise the end user prices a bit? So then the question is, will end users accept higher prices? Well, they don't accept higher prices. They're getting higher prices now because scalpers are buying everything up and reselling it to them. So yeah, we are definitely underpaying the entire manufacturing chain. Well, maybe my question is, do you think solutions like this will work out in the longer term or will the market just settle back in what it was before because that's the cheapest way to operate? Yeah, so the only long term solution is basically unfuck the system and optimize for stability rather than maximum profit. Okay, and the next question, microphone in the back please. Yeah, you told some insights about the car manufacturing industry and their supposed and maybe not shared by all practices of lean production. At least for the German car manufacturers, I also read that they had a problem because they outsource a lot of their stuff as components by third party vendors and that made it even more difficult to change something because they just get the part and then somebody else has to do that. Can you say, do you know anything more about that? Can you say something? Yeah, so what the issue there is basically those vendors are committed to a fixed price but their own costs go up passively. So they are just unable to deliver that price until basically their customers agree to pay more. I'm really sorry but we are out of time but I'm sure Clement will be available for questions afterwards. So please give him a nice round of applause.