E716 | This Week in European Tech with Dan Bowyer, Lomax Ward & Harry Destecroix

26:49Lomax Ward:loving it it's funny how everything is now moving into the real world love it yeah the only question I sort of that will time over to hell time will tell and this is an age old question in tech is do you take 100 billion do you buy a bunch of old economy manufacturing companies and juice them with ai and robotics or do you just start from scratch and go like we're going to build the airbus of the future we're going to build you know what i mean the defense cup we're going to build the rheinmetall of the future it's got to be that it's the only way to scale is that right it's got to be that i can't see any other way of doing this this strategy unless there is absolute, this is how we build tanks.

27:26Lomax Ward:This is how we build taps. I don't know what the things are, but I think this is going to be... He seems to be going the former route. He seems to be buying incumbents and juicing them rather than actually going greenfield and from scratch. Maybe it's simpler just to get around EU's planning laws. I mean, the factors, they own the property, right? I mean, it's just a way to say, here's a factory next to a train line. they have ip they have everything you know it's just like yeah brand maybe it's you know you've got the brand you've got the physical location and then you just put you know rebuild the whole thing with that and i think that's what the plan is i think it is to take in their full stack right so maybe it's the way of both those things happening simultaneously quicker if you've got like a certified thing like an aircraft if you fundamentally rip out and change the manufacturing process you almost have to like recertify the thing anyway sometimes i wonder you're better off just starting from scratch but look i mean these are smart people who can figure it out and you know but i'm sure time will tell and that there are we know there are full stack companies going after new i mean look at helsing i know helsing is doing some m &a but they're going after you know they're going off the defense sector in in europe um like from scratch i want to talk about another model so bezos is bringing here is that the chinese are bringing their ai token economy to market.

28:47Lomax Ward:So the way that I've described this in the descriptors, a new OPEC measured in tokens, not barrels. So China has formally adopted the Xi, I'm going to say this wrong, excuse me, Chinese people, Xi Yuan, as the Mandarin term for token at the 2026 China Development Forum. What they're doing is they're signaling the state of token-based AI economies. Now let me, This isn't crypto tokens, by the way, before people think this is some kind of crypto-ish thing. This isn't. This is about selling inference at scale because they own the whole stack. They own energy up all the way through. And because of the way that the country is run, they can effectively sell inference as a service.

29:28Lomax Ward:Now, Chinese models are priced at about 0.02 cents per million tokens, which is about 1.180th of the OpenAI price. The Chinese models occupy four of the top five positions on Opelruta globally, accounting for nearly 86 % of the total. So they're pushing this inference cost race to zero, which I think is something they can do because they own the whole bloody lot. I also think that Kursa has shown the world the way to go. So this week, and we're going to talk about this again in a second, but they're a$50 billion company who chose a seven-week-old Chinese open-source model over the U.S. alternatives because it was eight times cheaper.

30:04Lomax Ward:So we also know that Andreessen, for example, has already shown in some of their data that 80 % of startups are using the open source model. So I guess this is a blended way of asking the room, do we believe that buying tokens at scale is possible? Is this something that is not maybe a data or security or other sovereign risk? Do you think this is even possible? Is this a model that we will see come to pass? Maybe Harry, do you want to kick this one off? well where to start yeah i think it's clearly going to be a part of the mix people are going to always want you know the low cost cheap solution so you know and especially startups which are undercapitalized they're going to they're going to want to do this and especially you know rappers like cursor who you know are you know fighting for better margins but i think ultimately if you're building your enterprise app on something it's you know these models suffer massive problems of prompt injections and all sorts of issues so it's security but it's also reliability it's great that you can keep switching from these routers but ultimately if you want something that's going to be reproducible you're going to want to like have lock in certain models and of course you can maybe pick certain models for this but at the moment i think we're still in this wanting the best model wanting the best output is still not quite there and you know for me personally i'm addicted to opus also you've got kind of the next kind of ruben distance coming from nvidia you've got obviously the acquisition of grok this is super exciting because you're then seeing kind of you know orders of magnitude in terms of what per token but what i'm massively excited about is this kind of at the moment we're stuck at this kind of 200 tokens a second which is an eternity when you know when you're using this stuff so you know with the grok acquisition and that being incorporated we should be seeing kind of token speeds of nearly 2000 and like that's going to be really exciting so it's going to be high quality models super super fast so there'll be certain workloads and certain companies that will prefer very very low cost and don't mind the speeds but i think for the real meaty stuff the real high value stuff the real you know intelligent stuff you're going to want the top models and you want to you want to have best speeds in order to kind of solve the hardest problems i think there's i'm not fully with you on the on the quality and the speed i think they're not that far behind i it just feels like the if you do the 80 20 you're going to get most of the market using some version of this low cost service whether it's this token economy or other chinese open source models whatever it might be it just feels like a really sensible smart thing for most businesses to do not you harry because you want that level yeah i think sometimes i'm we're biased because we're we just want you know we just want to be using kind of the latest you want to geek out right yeah and that's a problem when you're kind of an early adopter i don't know if you're the cto of like a big healthcare organization insurance organized like i don't know like you you know well what's what what's the underlying model here i i don't know if that if that really works yeah no i mean look at microsoft i mean microsoft you know they've terrible software but because enterprises use them and you know everybody you get everything bundled in everyone has to use copilot only decent stuff they've ever built apart from i would argue exchange their initial stuff they've bought right what what have they actually built microsoft in decades all of their good stuff that people like they bought i'm i'm being a bit obtuse and rude there but anyway i'm gonna i don't want to go off topic lomax what other bits in here would you pick up on i think the the price of tokens is something maybe at the moment the people in in technology know fast forward a couple of years it will be kind of lingua franca on on the street like everyone will know it it will be like as tokens are becoming a commodity tokens are becoming something that everyone will think about because it's almost replacing a kind of unit of manpower or if you were taking an extreme example it's becoming the most important thing on the planet and it will only continue to be that so this formal adoption by the chinese of this term token i think is a you know important moment or it signifies how important these tokens are becoming.

34:13You know, as Jensen said recently, GTC tokens are the new commodity. And he's obviously working, as Harry said, on making sure that NVIDIA is competitive here. But it's, you know, this is the sign of the times, guys. You know, we talked about how the Chinese models are hugely widely adopted. They are going to have challenges, I think, for Western businesses that want to use them in enterprise, which is obviously a massive, massive market.

34:35Lomax Ward:I mean, if you look at Cursor as an example, obviously they are not i think reading behind the headlines behind the headlines it isn't their composer too which let's talk about now it was it was next on the docket but we can go straight into it it actually wasn't a full moonshot ai kimmy 2.5 ripoff i think maybe 20 30 percent of it was and they used their own models for the rest but they have been absolutely hammered for not coming up and saying look we built this on the infrastructure on moonshots kimmy kimmy 2.5 so i think it was the fact that it wasn't open or openly discussed that was the challenge rather than the technology that was used but doesn't that prove to your point low max this stuff is there and is is high enough quality and it is fast enough but it's definitely fast enough is it secure and safe enough right i think is the is the biggest i think you can mitigate that i think you can mitigate that well i don't know there is just prompt injecting stuff isn't there low max but you know nist in in the u.s found that chinese models are 12 times more susceptible to agent hijacking attacks so So, you know, and this stuff kind of resonates amongst CTOs and in big buying organizations.

35:40So look, I personally think of Thomas Tungens has a really good, always writes well, but has a really good post on this from this week. Actually, you know, this is this is a really cool thing for startups because, you know, you can use these open source models to build amazing products at a fraction of the price of, you know, you have to go and raise 100 million for compute now. So I think that is pretty cool. It might give you an enterprise sale ceiling, but, you know, smart entrepreneurs will find a way around. I mean, at the end of the day, once you've built your products and everybody's excited by it, you're just going to run these models in a European or American data center, right?

36:12So the advantage here is really if you ran the money in China, but you're not going to run them up. You know, if it's a European or an American, they're going to be running a data center in Europe or America. So, you know, NVIDIA still wins.

36:29Lomax Ward:Well, yeah. Yeah, thank you so much for that beautiful segue there, Harry. So ARM has built its own chip. After, what, 36 years of licensing chip designs, ARM is now building it. So now ARM is pretty much a softbank company, which I think is semi-problematic. ARM is the UK Cambridge-based firm. I'm pretty sure softbank own like 96 % of it now. They have released their AGI CPU, which is a production-ready chip built for running inference. which you've spoken about a lot in AI data centers, and Meta is its first customer. I'm pretty sure OpenAI is in there somewhere as well, but I'll have to look that up.

37:08Lomax Ward:So this is a good thing, more inference chip providers. So my question to the room is, how does, to your point, Harry, we don't need, I mean, NVIDIA is bringing down its own inference stuff and stock as well, but we don't need, you could run Chinese OS, you can run all these other chipsets from other manufacturers. is so how is this all gonna how is all this gonna wash through like do you want to set up some more of the more of the scene on this one this is an interesting one um which has been very very consistent consistent business model now is seeing 35 year old business model being sacrificed on the altar of ai as one commentator put it so now they're building their own cpu for inference jumping on the inference bandwagon they're like well you know why can't we have a piece of this which is the future right do we do we still agree that inference is the direction of travel from now I think for sure.

37:58I mean, I think, and this also gets a little bit uncomfortable for Arm because now they're competing with their own licensees, their own customers, right? NVIDIA, Qualcomm, others who pay them royalties. So I think that's something they've clearly managed to get their head around. But, you know, Arm is currently a$4.5 billion business growing at about 20, 25 % a year, 90%, maybe even 95 % gross margin, clearly, because it's a licensing business model. but you know maybe they're like 25 growth a year four and a half billion top line isn't enough i

38:25Lomax Ward:want a piece of this i want a piece of the i'm sure i'm sure soft bank is is gently on the phone going what the fuck guys can we get some more and you know actually i'm this what i haven't actually this will require a significant investment from them and this has been coming down this is not an overnight announcement this is this has been worked on for the last few years is you know this could take top line that that business line in itself could be many multiples of the existing business lines so obviously it will be less profitable but it's what it's more it's more a case now of who isn't building their own chip and also i think we're going to end up with who isn't generating their own power all of the all of the hyperscalers seem to be following the chinese model and trying to build their own full stacks which i think is a really really interesting thought experiment harry have you got anything that you wanted to poke in this one yeah it's it's strange so like agreeing with low mats like for them to sort of go away from that business model and there seems to be a lot of competition for what they're doing and of course it's more than just a you know it's an imprint cpu you've got all the networking you've got all of the other stuff around the data center so that they're they are they going to be doing anything else around that does this sort of threaten intel amd more if it's on the cpu side is intel still around they still they still got a business they've bounced back a little bit but trump's taking a stake in them well the u.s government has well i think they gifted they gifted 10 didn't they Please, please help us.

39:48Lomax Ward:Please don't shut us down. Please, please look after us. Right. Well, last on the docket. Actually, well, before we get to our predictions, our new section, I want to talk about Innovate UK. So they're getting a strategic overhaul. A founder is now running the show at last. But will he be able to do what's needed? So first up, for those that don't know, what is Innovate UK? It's a government funding funnel for R &D predominantly, part of the UK's mechanism for de-risking and leveraging private investment. I was working out where it fit into the government stack. So you've got Treasury, HMT, goes into DSIT, Department of Science, Innovation and Technology, into UKRI, UK Research and Investment, into Innovate UK, which is one of nine councils.

40:31Lomax Ward:So that's the flow of bureaucracy behind the scenes. UK has, I think it's 25 billion kind of R &D budget-ish. ukri is about 10 of that and then innovates it was beis wasn't it beis became dsit and two other quangos they can't keep it the same they have to constantly read relabel everything but yeah effectively ukri is the body that kind of funds fundamental research universities you then got all the different councils like biology council eprc physics so it's everything from quantum to bio to health so that all comes out of it's like university budgets plus infrastructure plus r &d spend and innovates billion is focused on innovative SMEs and firms that are kind of taking that technology.

41:20Lomax Ward:Right. And a year ago, exit entrepreneur Tom Adiola took the reins. Now, I think you've already said this, Harry, but he's allocating about a billion a year via grants, loans, knowledge transfer partnerships and computer access. They've had their challenges. Most people will remember the Women in Innovation fiasco 2024. They had a 75K award awarded to 50 women. They backtracked, only awarded 25. Two mil fell into an abyss, then apologized to the 1 ,500 applicants, and it was an unholy mess. It's obviously been noted as being slow, bureaucratic, often seen as a bit of a burden. Unreasonable computer says no responses to many of the requests and requests for support.

42:01Lomax Ward:And Adeola is now in charge, wants to shift from being a funding project into supporting more companies, make it more innovative, make it simpler and easier for companies to work with him so can he succeed should governments try to pick winners can a founder mindset actually work in this environment or will the civil service machine grind him down harry you know this world how would you how would you tee this up and where would you go yeah yeah for a full disclosure i'm on the innovate um uk council and and actually i i knew tom before he took up the post and he messaged me and um you know invite invited me to kind of I suppose they're the inner circle to kind of help steer this.

42:37So first of all, Tom's fantastic. And I think at the end of the day, founders coming into organizations which remits are to fund via grants, founders, it kind of makes sense to have some of that experience sitting in. So it's nice to see Chief Executive of a major double service department having a kind of a founder background. And I think Tom is definitely taking a fresh approach.

43:02Lomax Ward:we're seeing more on we've got alex to pledge in there we've had obviously matt clifford going in and out we're seeing more founders and entrepreneurs touch touching deeper into government and hopefully that will continue and hey they even invited me so i think absolute mistake but i think first of all we know i know a lot about this area i mean actually i was a recipient of an early invite k grant it was critical to kind of you know xylo's success story so they can be really really useful things i think the problem is like you've got the money's been very much spread very very thin tens of thousands of grants and then with ai you're just getting this like grant slot so that they're having to kind of change for a number of different reasons first of all i think it's they were very broad and i think now they're focusing around the industrial strategy they're focusing around there's an alignment now isn't there we've all kind of all been kicked into gear i think a little bit with the eight industrial strategies yeah it's like hey we invented the computer yet we've got like no computing companies right like the uk is you know unbelievably good at Nobel Prizes and fundamental research and unbelievably good at giving it to the rest of the world to commercialize.

44:06So I think like the focus around advanced technologies, quantum, semis, EngBio, for instance, is great because actually then what's happening is more of it's focusing towards our strengths and hopefully we're funding less, what I would call iterative R &D of kind of old medium-sized firms that are kind of getting, you know, innovate uk with some r &d tax credits and they're really they would have done it anyway so i think it's good to focus on what on what we're good at so that we can actually get that up and then it's also just to have this kind of thing you know the other thing i think is bringing in and i was you know there the other day and they mentioned the word deal flow and i was just like to hear that kind of in a government agency was great you know the organization needs to actually track the best companies why because you need to know what good looks like so you can benchmark everything and two hey by going to a high growth company and actually supporting it is it more likely to set up manufacturing in the UK do more development if it does raise that 200 million series b round is that now going to kind of be twice as many jobs in the UK and therefore you know the exchequer collecting that kind of tax receipts and it's less likely then if it does get acquired to keep a strong base in the UK.

45:24So I think, and at the same time, are they more likely to get that escape velocity? Because as we know, we've got to respect the power law. Most companies don't do very well. And, you know, the one or two companies in our portfolio deliver, you know, 90 % of the returns. And I think just understanding that and, and setting up the system is really exciting in terms of how hard it's going to be to implement. I think, you know, the thinking and speaking to all the right people but the proof's going to be in the execution and then finally i think one of the simplest things which you know i said to do is like most important thing i think for us is we don't want to waste founders time and there's lots of calls with really really high failure rates and biomedical catalysts for you know for health care is absolutely abysmal and you you really want to make sure that you get the screening criteria right and you get to know really really quickly you know we know that's really really important in in our trade is that you know if you mess a founder about they end up spending you know months in dd and then you you pull out the last minute that's you know that's actually net negative to that company so i think this idea of getting to quick yeses and nos and actually having having a prepared mind so you know what you're looking for is a good thing and they're going to have to do this because they're going to need to actually assess the companies on fundamentals because of ai slot and grant writing so you know it's going to be hard but i think you know i i back tom it's down to their execution at the end of the day but what i've seen is you know tom works at tom works at pace he's a proper founder and it's um it's going to be interesting to see how the civil service coats with that but there's definitely do you think do you think he'll be given the latitude

47:02Lomax Ward:that's what i feel is the challenge here will he be given the space and latitude i think um look founders are really good at kind of bashing their head against a wall until they get through right so if anyone's going to kind of make change it's going to be it's going to be it's going to be a great founder um that they're used to kind of overcoming big big challenges we need to get the funding up we need to get the funding up across the stack and this is part of the mix this won't solve it but it's you know it's it's going to be a multi-pronged affair isn't it harry it's going to be multi it's going to be bb over here a bit of bit of grant over here a bit of more private encouragement over there hopefully some pension poking up here so that whole pot will hopefully hopefully come to par and and it's people like us it's founders like you know you can sit and be an armchair critic but get involved yeah absolutely be supportive because i think that you know i think a lot of the interactions i've had has been you know they've been open ears and um it's great to see some of those changes that i you know in in various areas i've been advising dsip for for a few years now seeing some of those things actually starting to actually get through you know you've you've also got to kind of roll your sleeves up i think sometimes and um go help out it's a lot of good people um but it's a hard hard thing to kind of go against civil service well we are called upside so i'm loving your attitude uh lomax before we move on anything you wanted to add on this one no i think it can only be a good thing it's a hard hard task but i wish them all the best we are going to try a new section in this wonderful podcast called predictions and we need a we need a theme tune or i'm going to hum a theme to you do do do predictions lomax what do you predict i just remember that we're called upside this is not a great um this is not a great prediction is it's you it's yours misery you're predicting misery are you 100 gold you're not gonna my my my thinking and is that the revenue is just not going to come fast enough to justify the capex of these hyperscalers you know the big five hyperscalers amazon microsoft google metro and oracle will spend over 602 billion on infrastructure this year that's like they won't know it's like they will they will they will dan and that's like 33 up from last year and so and i just think you know to put that in perspective it's just the amazon's capex is alone is greater than that of the entire u.s energy sector it's just um insane so i think you know between now and 2030 there's sort of two trillion of ai related capex and i think if you look at where the revenue is today you know you total open ai anthropic and enterprise ai i don't know at best you're at a hundred billion and then you know you take that i think double that i think i'm pretty sure open i was looking at 150 by 2030 or something like that what's the limits you hear i was one of my friends is a hedge fund manager and he was saying that a lot of this has been um accelerated by this kind of depreciation rules that trump brought in effectively they can they can depreciate 100 of this kind of on day one so if you if you're high cash flow your effect is effectively taxed back and i don't i don't know how much you know that's actually hugely accelerating this infrastructure build out because you've got this kind of 100 depreciation like accounting thing that they've brought in and i i think the thing that breaks it though is yeah like you say is iran and and maybe that you know the saudis like not being able to participate i can't see how i can't there's just too much friction here so what's the prediction the prediction is revenue is not going to move fast enough to justify the capex is that the playback just won't be there especially when they need to keep ripping out these gpus every two years now putting them back in it's just like it's just um so people will realize that and people will realize they're over their skis and we'll have a general cooling off you know it's It's not a revolutionary prediction, but I think it's just still every time I go through numbers.

50:52Lomax Ward:Well, it's our first week of predictions. I think it's a very solid prediction. Yeah, but when I go through these numbers, I know Mads is not here this week. He's much more of an optimist than me, but I just think we're getting over our skis. He's coming more down our path. Did you notice last week he's coming down more this way? Hang on a minute. We are now moving out of free cash flow and into debt. We've got to think this stuff through. Harry, what was your prediction? i think i think you're you're seeing some of that heat coming out of the market that you know this the last month or so you've seen a real pullback you know across the board and a big rotation but i think what i'm seeing at the moment if you look at the 52 week prices on the kind of the mag 7 like you've just seen this kind of huge rotation you know microsoft is pretty much down it you know the lows on its 52-week range, so is Meta.

51:43But Google is still, you know, is doing strong. Apple's doing really, really strong. So I kind of feel like we're seeing the disruption of the Mag 7. And then you've got the IPOs of SpaceX, Anthropic, OpenAI. So I kind of feel like Microsoft Meta potentially going to be displaced. I think potentially if those two things come in, like you're going to get this kind of rotation into in these other stocks and you know are you going to see like this you know fragmentation i i'm kind of you know i to be fair in april i reduced my positions in microsoft and and meta just for these reasons because i just feel like you know they don't really have a mo or as much of a mo as i think say look at google google's all the way through from the best one of the best ai teams in the world it's demis you know demis is god um all the way through to kind of the hardware they're you know training gemini and the first tpu and they have the strongest distribution i mean by country mile i would argue and then you've got um you know apple and you've you you've got this kind of m5 kind of chips and being able to do you know inference on your actual device i won't talk about tesla because i'm not including it i'm

52:56Lomax Ward:very happy not not talking about elon that's fine with me yeah i saw that um that siri's now going to aggregate whichever model you want to use so they've they've culled their arrangement their unique arrangement with open ai which i think is the way to go i thought they were going to get fully into bed with the on-device gemini but we'll see maybe maybe they've worked out their own models i imagine so on the on the five and the last thing sorry is all those ipos probably won't happen if we this is a ran thing doesn't doesn't sort itself out why are you going to ipo if you You know, and I think if things are...

53:31Lomax Ward:Inflation hits 4 % in the States, which it's predicted to do. Yeah, that's going to be crushing for equities. So you're seeing that already, right? I mean, just looking at today, sort of 2 % down across the board. So last week's been all in the red. So yeah, don't know where that's going to go. Just to be extremely dull, my prediction is in exactly the same camp as both of yours. So my prediction, what I found fascinating this week is I was looking at how the hyperscalers are projecting revenue. So I was just picking on Anthropic just as an example, because there was a big discussion this week about how they did$6 billion in February.

54:11Lomax Ward:I don't know if you saw this. And I was like, they didn't do$6 billion in February. And they actually didn't. What they've done is they've got$6 billion added to their run rate, which basically, if it was$6 billion in February, they would be predicting$72 billion in ARR. So it's not. It's$6 billion to the run rate. It's around 20 bill. We're doing extremely well. I mean, we're still talking buckets of cash here. I'm not trying to be demeaning in any way. However, how they predict and project their revenue is they take usage spikes from that month and then forward plan 12 months. So it's all these people experimenting and spending gazillions on tokens.

54:51Lomax Ward:And then I was looking at the, what other fact did I read? It was eight of the 10, the Fortune 10 companies are all Anthropik customers. But you don't have to scratch a little bit below the surface to see that there, some of them are the low cost, low spend. Okay, you got somebody in the eight of the 10 using the tech, but it's just not the volume that people think. there's going to be i think a reckoning i think we're going to start seeing how how these nonsense capex spends where i'm not with you low max i think a lot of this capex just can't happen just throw energy and war into the pot and they can't happen let alone planning and all the other frictions that we're going to see so i think people are going to my prediction is that people are going to start to wake up to how these how these revenue lines are pushed into the press and how they're absolutely nonsense.

55:38Lomax Ward:Last thing before we close off on this. Interestingly, 79 % of OpenAI customers also pay for Anthropic products. So everybody's buying everything and trialing everything. And I think that's just going to stop. I think you're going to start, I think obviously I'm not a very good representative of the general population, but I think you're starting to sort of just, you know, you've been trying everything. You're starting to sort of get your kind of tech stack locked in now. And I've been, you know, got rid of OpenAI. a while back now Anthropic curse has now gone so you're starting to kind of it's really like you know Anthropic and Google for me the other thing I thought was interesting is kind of how they set their prices and forecasts it sounds very much like how the energy markets in the UK work where you know 10-20 % of let's just set up the gas price and that's what we should all get paid anyway yeah I'm going to go with that deal of the week Lomax quickie deal of the week deal of the week is Well, Nathan, AirStreet is a fund.

56:39Yeah, good for him. Good for him. His solo GP raised, AirStreet 3, I think it is, is$232 billion, a million? Million, yeah. So good on him. He's been long on AI ever since he set up the funds. So good for Nathan. It's great to see kind of new GPs coming in and raising that kind of capital.

57:00Lomax Ward:He's solo as well, right? Yeah. Mine was Granola. I love the fact that Granola becomes the latest UK unicorn. I love Granola. I think it's a bloody awesome toolkit. Harry, did you have a deal of the week? I think mine was this lace lithography. I can never say lithography. Everyone's got one of those words. Funny enough, we've looked at a few of these kind of companies. I won't say that word again. But yeah, effectively, what they're trying to do is get below this kind of EUV, kind of 13 and a half nanometer limit so effectively when you make chips and you put the designs into silicon you need a really really small short wavelength of light to make these tiny tiny features and effectively we've kind of Moore's law in terms of transistors has stopped now because you know we can't get past this kind of sub nanometer resolution at the transistor level I wanted to send this to Alex in our team who's got a PhD in optical physics but I didn't get to So I've done my best to try and explain it.

58:02But yeah, interesting, great. You know, whether it's a competitor to ASML.

58:06Lomax Ward:What was this, Norwegian? Where are they? Where's Lester Thornton? Norwegian Deep Tech Startup Company. Gentlemen, loving that. Loved our chat. Thank you so much. And I'll catch you all next week.

58:30Thank you.