The Turning Point: The Story of ASML

0:00I'm Jonathan Ford and if you like financial history as much as we do on a long time in finance we've got some excellent news. Our new friends, the Library of Mistakes, run a brilliant course on advanced valuation in financial markets, written and presented by leading academics, senior economists and investment practitioners and hosted by one of the UK's leading business schools. The course covers key areas, including equity valuations, economics, behavioural finance and liquidity. And the really good news, they're giving listeners to the podcast a 25 % discount. Just go to the link in the show notes and when you get to the booking page, enter ALTIF25.

0:42That's A-L-T-I-F 25 in the box marked promo code. Also, we have some other excellent news. Yep, we're doing a live gig, listeners. It's an unmissable chance to see me and Neil in person at the Weekend of Mistakes, a tremendous festival of economic and financial history in the beautiful Welsh town of Hay-on-Wye between the 20th and the 22nd of March. Now, we'll be doing a live podcast on the 21st, that's the Saturday, where we'll be talking about the art of the squeal. Drawing on our decades of experience and more than a couple of colourful war stories, we'll unpack how supposedly sophisticated lenders keep getting squeezed and how entrepreneurs and investors can use restructurings and workouts to avoid paying what they owe.

1:34Now to get tickets for this prestigious event, go to weekendofmistakes.org and get tickets. And there are even special rates for Altif subscribers on regular and premium tickets. You just need to book with the code altif underscore offer. That's A-L-T-I-F underscore offer.

1:58Thanks for listening to A Long Time in Finance with Jonathan Ford and Neil Collins.

2:08Welcome to The Turning Point, a new series from a long time in finance. Now in each episode, we're going to look at a key event that changed the course of a company or an industry's history and transformed it either for good or ill. We'll discuss topics such as why the company that invented the digital camera was destroyed by it, and how Bernard Arnault turned a small construction company in northern France into the world's largest luxury goods business, essentially through one brilliant business deal. They all show how chance, individual agency, a moment of inspiration, as much as wider economic or financial forces, can drive the fortunes of companies or entire industries.

2:49Now, in this episode, we're going to talk about ASML, a company you may never have heard of, but is actually one of Europe's largest companies and by far its largest tech company. It's worth around 450 billion euro, twice as much as the second company, SAP, a German software company. Founded in 1984 and headquartered near Eindhoven in the Netherlands, ASML is up there with Silicon Valley's finest. But the strange thing about this story is that ASML didn't devise the technology that made its great fortune, nor did any researchers in the Netherlands. Other companies, mainly in America, were responsible.

3:32So how did this unbelievably lucrative technology used to imprint circuits on silicon wafers to drive the chips that run everything from iPhones to cars, or even the fridge in your home, end up in a sleepy Dutch town called Veldhoven? And here to help us unravel the mystery is an old friend of the show, Brian Potter, Chief Infrastructure Fellow from the Institute for Progress in Washington. Welcome back, Brian. Thanks for having me. But first of all, we should start by talking a little bit about the industry ASML is in and the problem that the technology it developed helped to solve, which I think is called Moore's Law.

4:10But maybe you can tell us a little bit about the background to the story we're going to tell. Sure. So ASML broadly is in the microchip manufacturing business. And microchips are made by etching extremely, extremely tiny patterns onto silicon wafers, you know, over in layers and layers of these patterns over and over again. ASML makes a series of machines called lithography machines, which are the machines that actually imprint those patterns onto the wafers themselves. It's one of the most important parts of semiconductor manufacturing. It's one of the most expensive parts of semiconductor manufacturing.

4:48in a new sort of semiconductor fab, the lithography machines will be something like 25 to 30 % of the cost of the entire fab. And a new fab will cost, you know, tens of billions of dollars to manufacture. So they make these lithography machines that sort of help etch the patterns onto these chips. That's essentially the only thing that they make. And they've, you know, become exceptionally skilled at manufacturing them. They make the machines that are by far the most advanced in the industry. Nobody else can sort of make the machines that are capable of etching patterns as tiny as they are capable of, which has given them, you know, an extremely important place in the industry because they can make the most advanced chips.

5:27Making the most advanced chips, most advanced semiconductors require these machines from ASML to produce. Okay. And essentially the history of this industry since whoever it was, William Shockley, I think was the American engineer who invented the first one, silicon chip, basically has revolved around shrinking or increasing the circuits that can be put on ever smaller wafers of silicon. And this is expressed in a thing called Moore's Law. Yeah. So there's a few key elements, I would say, in the history of semiconductors. One is the invention of the transistor, which Shockley and a couple other guys at Bell Labs, they built the first transistor out of this sort of solid state switch and amplifier that could do jobs that previously were done by vacuum tubes, but use much less power and be made much, much smaller.

6:17And then over time, these transistors got smaller and smaller. And then they realized that previously, you know, the first transistors were each transistor was its own little discrete component. And then they eventually figured out that you could make a bunch of transistors all at once on a sort of single silicon wafer. And that was, yeah, that was another sort of key invention, these sort of integrated circuits. By the sort of early 1960s, this sort of basic technology was in place where you could make these integrated circuits that consisted of lots and lots of semiconductor components all on sort of one single wafer.

6:49And you could make these in a very repetitive, very efficient process. So those things together were sort of the key that made it possible to kind of this technology to continue to advance. And then in the 1960s, yeah, a guy, Gordon Moore at a company named Intel, noticed that the sort of number of components on these integrated circuits was increasing fairly consistently. every i think at the time it was a year and a half and it was later extended to two years but every two years the number of components on these chips was uh doubling and then over time that just doubling continued and continued and continued and so as transistors have gotten smaller and smaller and they've been able to cram more and more components on these uh chips has been a sort of major vehicle for how this technology has advanced so the way they actually imprint these circuits because as you say they become incredibly small is by shining beams of light or It's like a process of taking a photograph initially, isn't it?

7:43What happens is you have a silicon wafer and you coat it with a material called photoresist. And then you have a mask, which just has the pattern that you want to sort of inscribe on your chip. And you shine light through this mask and it exposes just certain parts of the wafer to light. And the parts where they are exposed to this light, the photoresist either hardens or softens, depending on the sort of type of photoresist you're using. and then they sort of wash away the sort of softened parts of the photoresist and that's left what is left behind is sort of a pattern that you want to etch in this photoresist and then you sort of expose this wafer to sort of an acid or a plasma that will then etch the pattern and then you sort of wash the photoresist away and you kind of repeat you know deposit a new layer of material and you kind of repeat that over and over again so basically we now bring it up i I suppose, to the problem which ultimately leads to ASML's opportunity, which is initially they're using, I think, ordinary light to shine onto the photoresist to imprint the circuits on the silicon wafers.

8:47But they reach a constraint as they shrink things down that at a certain point, ordinary light doesn't work so well. And it's harder and harder to get the precision that they need to imprint the circuits? So the problem that you run into is that when you shine light through an opening, like a pattern in a mask that you're trying to sort of, you know, imprint on this chip, whenever it goes through an opening, it kind of at the edges, it kind of wants to spread out a little bit. Diffraction. Yeah. And the larger the wavelength of light that you're using, the more that you have this diffraction problem.

9:23And so if you want to sort of continue to shrink your features and it's like, you know, smaller and smaller and smaller patterns in your semiconductors and keep up with Moore's law, basically, you eventually run into sort of diffraction limits at sort of a particular, you know, wavelength, right? You'll run into a sort of barrier that you can't overcome at some wavelength. A way to overcome that is to use shorter and shorter wavelengths. For a long time, they were able to just do like yet visible, normal sort of visible light. And they found kind of various clever ways of sort of overcoming these diffraction limits.

10:00But eventually there's only so many clever things that you can do. And eventually you just run into these fundamental physical limits. And so you need to sort of switch to a shorter wavelength of light. And they'd done that a few times over the course of the sort of semiconductor industry, I think. Historically, it was done with like 435 nanometer wavelength light. They had kind of shrunk that down to sort of 193, I believe it is, nanometer light. But they needed to find ways to sort of shrink that even more to sort of, again, overcome these diffraction limits, be able to kind of continue to etch smaller and smaller patterns.

10:33The extraordinary thing is they obviously start thinking about this a long time in advance. So people start looking at ways of doing things differently back in the 60s. Yeah, very early on, they saw that these features were getting like smaller and smaller and smaller, right? And they realized, well, eventually we're going to run into these kind of diffraction limits. They thought the limits to sort of what's called optical lithography, which is, you know, using like visible light, like normal light. They thought those limits would come much earlier. There's a saying, there's a sort of saying in the semiconductor industry, the end of optical lithography is always six or seven years away.

11:05They just keep finding ways to like... But eventually, like I said, you kind of run out of these sort of tricks and eventually you need to sort of switch to a narrower wavelength of light to be able to kind of... Yeah, shorter. Yeah, advancing on Morse law and shrinking your components. So they look at two options. One is they look at electrons and the second is they look at using x-rays. Ultimately, they go with the latter. There was actually quite a few different technologies they considered. I think at one point in the 90s, there were sort of four different ones that they were considering. One was, yeah, x-ray lithography, which is, yeah, using x-rays, which are between like, I think, 0.2 and 20 nanometers of light.

11:45There was, yeah, electrons. And there was another one called ion etching. And then, yeah, this last one, what was at the time called soft x-rays. But then later they changed the name of it to a UV. For a long period of time, up through the 1990s, EUV did not look especially promising. People thought it was going to be x-rays were going to be sort of the next big thing. And all these sort of chip companies invested all sorts of money and development effort into trying to make x-ray lithography work in sort of the 80s and the 90s. And it just was very hard to kind of make it practical. Yeah, the further you go up the electromagnetic spectrum, the harder it is to deal with the waves that are produced.

12:30That's the point, isn't it? Yeah, as I understand it, that's a big problem because as you go into these shorter and shorter wavelengths, you run into things like, you know, with normal light, you can just like bend it with a lens or like reflect it with a mirror. And so you can like, you know, concentrate these light waves into sort of this very small patterns that you need. It turns out to be a lot harder to do with like these shorter wavelengths with things like x-rays. X-rays want to just mostly like go through stuff instead of being like focused with a lens or reflected by a mirror or stuff like that.

13:00So you run into sort of all these very gnarly technical problems when you start going into these shorter and shorter wavelengths. Let's talk a bit about how this solution is discovered, because I think it starts in the 80s in Japan, doesn't it? a guy with the name of Hiru Kinoshita, who is a researcher at NTT, the telephone company. He's playing around with these ideas, but he's playing around with x-ray lithography, and he starts to develop the idea of soft x-rays, which maybe you can explain to us what they are. Like you said, he's at NTT, he's in Japan working on x-ray lithography, and x-ray lithography was done using this method called like proximity lithography.

13:41You know, with normal like optical lithography, you would shine your light through like a lens to like focus it down to sort of a small point. But with x-ray lithography, as I understand it, they didn't know really how to build a lens to sort of focus these x-rays. So you just had to shine it directly on the mask. And that sort of ran into various difficulties. If a mask had any sort of defects into it, if a normal like optical lithography those defects would kind of be shrunk as you sort of focus the light down but with x-ray proximity lithography those defects wouldn't get shrunk and so the tolerance for like mass fabrication was like much much much tighter and much more difficult to achieve as i understand it that was one of the issues but they're having all these issues with sort of x-ray lithography and he sort of started to look to see if he could do it with you know yeah what were called soft x-rays which were like slightly longer wavelengths of electromagnetic magnetic radiation.

14:34And he sort of learned that there's had been some research previously that with these soft x-rays, you couldn't focus them with a lens, but you could bend them with a very special mirror, which is this thing called a multi-layer mirror, which is made of these like lots and lots of layers. So like this extremely precise, precisely laid down material. And that does something called like constructive interference, which is sort of part of the x-ray radiation gets like bounced back and it gets bounced back in such a way that each little layer it reflects a piece of it and all those sort of reflections kind of add up and so yeah he started looking into it in sort of the late 1980s he kind of gets some very early successes and he sort of starts talking about his work and many many people are like very skeptical of this idea they don't think it will ever be practical but a couple people elsewhere started like oh maybe that would kind of be a interesting idea and to kind of see guys at uh bell labs in the u.s kind of like pick up these ideas and start looking at it a little bit.

15:31So really what we're talking about is not, you know, there are lots of ideas being kicked around. It's not that they don't work. Most of them can be seen to work in some way or other. It's just a question of, can they be commercialized as some sort of kind of mass production model? Can you actually have these mirrors working in a factory environment where they're churning out lots of wafers? Yeah, exactly. That ends up being One of the main sticking points is that a lot of these things, you can get it to work and fabricate chips with it, etch these patterns. But there turns out to be a very large difference between getting it to work and getting it to work in a commercial mass production environment.

16:10Well, you talked earlier, you talked about EUV, which I will now reveal. It stands for extreme ultraviolet lithography. this is essentially the process that gets developed out of the Japanese researchers ideas as you say I think Bell pick it up in the United States and essentially the question is who's going to turn it into a commercial product if it's going to be turned into a commercial product and that really brings us to the 1990s so it starts to get a little bit more momentum Intel kicked some money into funding it in the early 1990s. The US government kicked some money into funding it. So they're kind of developing this technology.

16:51It still does not seem like it's obviously going to be the winner. I think as late as like the mid to late 90s, EUV was still maybe not considered amazingly promising. But it was promising enough that it was worth investing this money into, you know, placing a bet on it. Basically, people start kicking money, it kind gains more momentum. And then sort of in the mid 90s, US government basically decides to stop funding the development of this technology. And so Intel ends up stepping in to sort of fill the gap. And they sort of organize this US consortium that various people contribute money to called EUV LLC, which is just sort of this like special purpose company that is going to like continue to work to develop this technology.

17:37The curious thing is that how come this technology was not actually developed to the commercial level that it is today in America rather than in an obscure European country? It's rather unfair to Holland. Yeah, it's actually quite interesting. So Intel was like the big company that was like pushing this development forward. Basically, they wanted the next generation of lithography technology to make better and smaller and faster chips and keep kind of keep Moore's laws going. And to do that, they wanted a major manufacturer of lithography equipment to sort of adopt it, because without a major manufacturer to adopt it, it wouldn't become the standard.

18:23And so they needed a major company. But the problem was kind of by the late 1990s, all the major lithography companies were these foreign companies, non-US companies. The two leading companies were Nikon and Canon, these Japanese companies. At the time, the US was very averse to letting the Japanese get their hands on this technology. The US semiconductor industry had basically just been nearly destroyed by Japan in the 1980s. I can remember that what they would do, I mean, it looked as though Japan was going to conquer the semiconductor world, along with the car manufacturing world, and everybody else would be sidelined.

19:04So there was a sort of great paranoia about the inexorable advance of Japan, which must have colored people's views in those days. Yeah, there was just no way that they were going to let the Japanese get this technology. But again, you couldn't really just given it's like American companies wasn't good enough that they had just such a tiny market share, they're getting increasingly squeezed out of the business. And so you had, you know, the two big companies, Reedge, Nikon and Canon. But then the third company was this Dutch company called ASML. They had like a 20%, I think, global market share in lithography market by the sort of mid 1990s.

19:40They were formed in 1984. They had sort of started as a joint venture between Philips, which is, you know, the big Dutch electronics conglomerate. And yeah, this company called ASM International. Which is a fascinating company. I tried to find out more about, it was all set up by a Dutch entrepreneur called Arthur Del Prado, who grew up after the Second World War, went over to Silicon Valley in the 50s, having thought about going to Harvard Business School, but then flunking out because he couldn't see much point in it. And he basically, of course, falls into the whole circle of William Shockley and all the people who set up Fairchild Semiconductor.

20:20That is kind of the origin of ASM, which is the company that leads on to ASML. So there's a kind of way in which Gordon Moore of Intel, one of the people who ran away from Fairchild, they're all sort of in some way linked back to that point in the late 50s in America. It's quite kind of fascinating. Also, to do him credit, you know, Del Prado's created three companies which still exist, are all quoted with a collective value of about half a billion euro. But there wasn't a single obituary of him in the Wall Street Journal or any of the British press that I could find when he died in 2016. Yeah, that's interesting.

20:55That's, yeah, disappointing. Yeah, he had formed this company, ASM International, which I believe made some sort of furnace for manufacturing these semiconductor chips. He really wanted to kind of get into lithography game. Phillips at the time was making its, like, lithography machines because at the time, anybody that made chips also made the machines to manufacture them. But by the 1980s, that was looking less compelling And people were trying to sort of, Phillips was trying to like get out of their lithography machine game. And so they formed this joint venture, they spun off their lithography efforts into this new company called ASM Lithography, which later became known as ASML.

21:32And they really struggle for like a few years. It's kind of a while before it looks like it will be a bet that is going to pay off because they sort of get spun off and they immediately need like a massive investment to sort of create the next generation of lithography machines that they need to compete in the market. But they sort of slowly start to gain more successes in the late 1980s. They were able to get in with TSMC, which ended up being quite important. By the 90s, they're selling to IBM and Samsung, and they've carved out a pretty large chunk of the market. And so, again, going back to the EUV, Intel kind of needs a foreign firm to sort of buy into this technology.

22:09And almost by process of elimination, ASML is sort of the only one that is acceptable to them. Yeah. So basically, Intel's presumably thinking is, we don't want a Japanese company to do it, because what happens if they support the Japanese industry and don't give us a fair shake? It was almost sort of anyone but the Japanese. And this rather obscure European company was the one that fitted the bill. Yeah. So yeah, they end up getting a license for it. And then a few years later, after they get this license, several US companies get these licenses as well. But again, And they're slowly getting pushed out of the market.

22:44And a few years later, ASML buys one of the last U.S. lithography companies that stand in this company called Silicon Valley Group SVG. And they did that so they could help them get land intel as a customer. So they end up buying SVG. And so now they're really like the last person standing. Well, at that moment, was it clear that that technology was going to be the one that emerges as the most important, seeing all the others off? Or was it just one of a number which might emerge? By the time that like sort of the deal goes through, the industry is starting to coalesce around EUV as the next standard.

23:25But it turns out, you know, even though they're increasingly like making their bets on EUV, but it ends up like just because they've gotten this technology like quite far through development and kind of by the late 1990s, the early 2000s, they can kind of show that it basically works. They can use it to sort of create these patterns on these silicon wafers. And presumably, having proved the concept, as it were, it was quite clear, having got the concept, that it was capable of much more development, whereas some of the competing ones were towards the end of the stretch that they could take. I mean, that brings on to the question of what were the others doing?

24:02Because there are several other firms in this industry, an American one called Ultratech, which decides not to pursue this EUV, even though they were involved in the consortium you talked about. And then what are Canon and Nikon doing? Are they just sticking with optical or are they trying other technologies to see whether they can make them work? There was this U.S. consortium, EUVLC, that funded the development of this technology. And there ended up being similar consortiums elsewhere. There was a European consortium called Euclides that was a similar effort. I think it may have later merged with the U.S.

24:38effort. And then there was a Japanese consortium as well. But those ended up able to take it across the finish line. And I think a big part of that was just, again, it was an incredibly risky thing to do. Even after it was demonstrated that you could make this thing work, it took an enormous amount of time and effort to get these machines actually practical working. And most of these other firms just decided they couldn't take the risk and spend the enormous sums to get this technology where it needed to be. So basically, the technology, they pick it up. As you say, they wrestle it over the finishing line, buying other businesses as they go to try and make it work.

25:18It's not until around 2013 that they actually start shipping these devices, isn't it? Yeah, starting in the 2010s, they start to ship them. But even when they ship them, they're still not really being used in actual production. It's not even until like several years later that they're really starting to, okay, we can like really, you can use this in actual factory for large volume production. It takes like a very, very long time. I think it might be 2017 or something when they first start to be like really used for volume production. It takes like a very, very long time. And huge, huge amounts of investment.

25:49It takes so much money that ASML has to request investment from the major chip manufacturers. And so TSMC and Samsung and Intel all end up investing in the company ASML in exchange. ASML uses that money to fund the continued development of this technology. So is it now the dominant technology? It is the dominant technology for like the leading edge, the most advanced semiconductors that need like the smallest and most advanced chip patterning. Older technologies are still quite widely used because a lot of chips don't require this like most advanced technology, right? You know, automotive chips and things like that.

26:28A lot of these can be done in like older process nodes that don't require these most advanced chips. But for the most advanced chips, it is pretty much the only game in town. And in terms of EUV, is ASML still the only show in town? Yeah, I believe there's only one. I think the Chinese are working on it, and I think they've been making progress. As far as I know, it's a certain production-ready scale in a large volume. Pretty much just ASML. Yeah, they're very profitable, these machines. So I was looking at the data point from 2019, which would have been early in the development or the rollout of this technology.

27:03So ASML in 2019 sold 229 lithography machines worldwide, of which 26 were EUV devices. But EUV devices accounted for a third of their revenue. And they're now, I think, thought to be as closer to about 75 % of their revenues come from EUV or maybe more. So it's obviously been a success from their perspective. Yeah, I'm sure it has. Yeah, they made a very, very bold bet. I mean, it's almost like a morality tale that it's actually, you know, you invest in something which some others might have thought was unpromising and what's wrong with what we've got. And in the end, you produce something which is clearly today in a league of its own and repaying that investment handsomely.

27:54Yeah, but of course, for every success like that, there's many that, you know, huge investments that don't pay off. I think in the 80s, we talked about x-ray lithography. IBM, which at the time was like a big manufacturer of chips and semiconductors, they invested like a billion dollars in, you know, 1980s dollars. A bold bet on x-ray, you know, lithography being the next advance in semiconductor manufacturing. And that ended up being wrong. And, you know, x-ray lithography can use it for some things, but it never sort of took over optical lithography as the main thing for, you know, these high volume mass production of chips.

28:29So, you know, they made a bold bet that turned out to be wrong, which is why it makes it bold, right? There's a very good chance that it won't work out. Yeah, yeah, yeah, yeah. You know, people in Britain often talk about the idea that we're very good at inventing ideas and then not commercializing them. This is an example of a Silicon Valley researchers, developers coming up with something and in the end it gets exploited by a Dutch company in this case, not by an American one. Do you think in America people feel that this one slipped through their fingers? Or do you think that basically everything worked out as they planned?

29:03You know, we didn't want the Japanese to get it. They didn't. I suppose the Dutch don't have a big chip manufacturing business because Philips got out of that years and years ago. So they're unconflicted in who they sell to. My sense is that the sort of disappointment around the technology is more about Intel and their missteps rather than ASML having the technology. I think for one, very few people are familiar with like, I think this history and the idea that like, oh, all this was like initially funded by U.S. companies and the U.S. government. But I think the sort of disappointment is more that, you know, Intel invested all this money and time and effort into getting this technology up and running and working and commercially viable.

29:44But then at sort of the crucial moment, they decide not to sort of take the next step with it in the 2010s. They decide to sort of take another step with like the previous optical lithography, because at the time it was like, again, didn't seem like it was quite ready. tsmc you know the taiwanese company they did basically take this bet on uh euv as the sort of next generation and that was part of what let tsmc leap ahead of intel how interesting i had a look at some market cap or capitalizations of various companies we've talked about so asml is now worth around 500 billion dollars intel is worth 234 so essentially the company it helped to create to supply it with equipment it doesn't actually use it's now twice as valuable it has now it is now it is now buying these euv machines or whatever it is it's it's basically of course corrected and is now fully on board euv but i think the sort of disappointment yeah around euv is more about intel losing ground to tsmc and euv's role in that than it is about asml getting the technology.

30:54Because yeah, by the time ASML gets it, the US sort of lithography industry is already fairly atrophied. It's not obvious the US companies could have taken it up to market and be successful. They certainly didn't have the funds to do it themselves. Intel, of course, is a cautionary tale of its own. I mean, in 2008, it bestrode the boom like a colossus. And then the price is now, I think today, about one-tenth, the share price of that one-tenth of what it was at the peak in 2008 and is sort of struggling with a little help from the federal government but it still looks like an also-ran. It's a different story but it's an interesting tale.

31:34Yeah and the other interesting thing I just tossed in about ASML, also interesting to me apart from the fact that its founder didn't get an obituary anywhere, if any, if he did by the way, any listener finds he did, do let me know, is that the company also, I mean literally I've never heard of any of the people who run it it's basically run by a bunch of dutch engineers who tend to serve for about five years and then move on to the next guy so it's the absolute reverse of the u.s sort of tech scene where everyone knows who the ceo of any company is yeah it's interesting i believe in like maybe seven connector circles the guys that run it are a little bit well known i think as of at least a couple years ago it had sort of not co-ceos but like it was led by like two guys one that's like very technical guy uh martin vandenbrink very talented, very sort of renowned engineer.

32:20And then another finance guy who's winning, I think his name is, but super not well known, like some of these like famous US CEOs are. Well, just to show the benefit of going with your gut and taking a risk, it's worth also just ending by looking at the market capitalization of the companies in the lithography market who didn't go with this technology or didn't make it work. Ultratech, I'm not even sure they're still in the lithography business, but they still exist. they're worth 823 million dollars million not billion nikon is worth 3.9 billion and canon is worth 29 billion so it's quite a extraordinary story really of a kind of different fortunes based around one bet where the american government in part and also american industry basically picked one company and didn't want the others yeah very interesting and of course much much that had to go right after that big turning point.

33:14But yeah, that certainly was quite important. That's the turning point phrase, which we've covered very well. That was A Long Time in Finance with Jonathan Ford and Neil Collins. Production and editing by Nick Hilton. If you enjoyed the show, please rate and review it on your podcast app, as that will help new listeners find us.

33:42Thank you.