E651 | This Week in European Tech: Exit Taxes, AI Reality Checks & The New Tech Sovereignty Race

0:00Well, hello and welcome to Upside, where we look behind the headlines affecting European venture. Today, Lomax, Mads and myself. And we are talking about, well, we're talking about very briefly some UK politics. So we are going to go into the exit, no exit, repealed, Reeves repealed tax. We're going to look at state of AI. We can look at the Einride SPAC, EU and Chinese vendors, a little on Palantir. That's in the quick news section. And then in the deeper dive, we are going to look at some tech sovereignty, the cyber attacks still piling in, Chinese kill switches the hyperscalers accounting fraud according to Michael Burry we'll we'll have a look at that now he's uh he's just shutting up shop and Mads I'm going to lean on you to dig into that latest in AI and deal of the week

1:04now i i can't not start without mentioning our chat last night gents so we were all chatting late into the night about how much we should shouldn't go into what's going on in uk politics the Reeves budget. And we must talk about the exit tax, which is now no longer. So I don't know, Mads, you were the most resistant to talk about this. And I know why. And it makes total sense. We need to just do heads down and build. It kind of almost doesn't affect us. It does. But come on, let's just get our engineering, efficiency, productivity, AI investing startup hats on. But tell us more. So can I go into a little bit of it very quickly, very quickly, promise.

1:51The idea of an exit tax is something that other countries in the world have introduced at various points in time. It turns out that as you keep raising tax rates, eventually people that are entrepreneurial and industrious and like to make things and build things eventually say, gosh, this is really a little bit expensive to be around here. I think I just might go somewhere else. And so to stem the exodus of entrepreneurs from some countries, they are saying, well, if you leave, we're going to tax unrealized gains as if you had sold all your stuff and you're going to have to kind of answer for any tax you might be liable for, even if it turns out at some point in the future that the thing you were building turns out not to be worth what you said it was.

2:41So how does that apply to venture? Well, let's say you're a founder who does really well and you build a business to, let's say,$100 million in value. And let's say after your Series B, you're still owning 20 % of that company. So you've taken something and turned it into$20 million. And let's say you now need to move to the U.S. to keep growing that business and making it into a global category leader. Well, if you had an exit tax, you would be taxed the day you left as if you had sold your assets and you'd be liable for that tax. And the way this is interpreted in some countries is even if the business then later fails and you end up not having any money or making any money out of it, you still have to pay tax on that notional gain you had the day you left.

3:31And that's, of course, a terrible idea. And it was mooted for about five seconds here in the UK. and because it's such a terrible idea, everybody got very excited about it and it was discussed a lot. But I think even the not always terribly business-minded Labour government could see just how incredibly dumb that idea is and suddenly they shelved it again. So I think that pretty much sums it up. Have I missed anything? No, I mean, I think it's also good to share. There was a lot of noise around it and I don't think people fully understood that it was an exit. as in you leaving the country tax rather than an exit leaving a business tax, although interconnected.

4:13And also, I don't think most people know that Germany and Norway have an exit tax. And so there is, so there is a, there is, and I understand, I think also people got on their high horse and didn't fully understand the, the intent of the government to say, look, we want that money to stay here and be recycled here and to build here. Not, not fully appreciating what you've just explained very well, Mads, which is, you know, it just doesn't work. And all people are going to do is go elsewhere. It's just a silly, pointless thing. Well, that's it. I mean, there's sort of, now we're on it, right? We have to go here.

4:46I mean, there's the other thing, which is the UK is not a very big country and the world is a big place. And the way we are going to win is by being a destination of choice for the best and the brightest founders to come here and set up business. Which we've done before, right? We've done this before. Which we absolutely, absolutely have, right? The UK was that London was Europe's number one talent magnet. And just imagine what a tax like this would do. It would absolutely say to talent everywhere, we're close for business, don't come here, we're not business friendly. And so it's just totally self-defeating and something that hopefully we can pull a stake through the heart of this devilishly thing and hope it never comes back.

5:30Well, she was definitely pitch rolling it well, wasn't she? Well, to be fair, it was pitch rolled for like, of five and a half seconds and then recounted the Reeves repeal was speedy. But I love the way that these notional kind of things are thrown out there. Oh, bad idea. Okay, we'll just take that one back. I think there is an element of deliberate, like some of this is deliberate. Historically, I think this has always happened, is that in the run-up to a budget, they selectively leak certain ideas that they have to test them. Do you think it's a blue elephant? Do you think it's purposeful other than hide something else under the carpet?

6:07I don't know if it's to hide anything. I think it's to test things. It's to test things to see how the markets would take them, to test things to see how they would be received politically. Because otherwise you end up, you know, if you think about all of the backfires over the years in UK politics, like the pasty tax, the granny tax, the whatever you want to call it tax, it's to avoid those kind of headlines, I think. but it goes back to the let's not spend a lot of time on this but i will just say it does go back to the point we won't there is a there is a feeling that the government don't know what it's they don't know what they're doing and and if you look at the budget last year and the budget this year have been significantly delayed and when you delay a budget you leave open a lot of room for conjecture like this so it's generally not healthy at all i think we need to get to the budget next week and then um and then move on in fact or 10 days time whatever it may be let's um let's move swiftly on i know i know a very unpopular topic but i didn't know how we couldn't not cover it the only thing that was interesting from tom tom bloomfield the founder of monzo proposed something on on social media which effectively is that um for the period in which if you did leave the uk and then sold your business subsequently for the period in which the gains accrued while you were in the UK, you would pay UK capital gains tax on that.

7:28And then, you know, in the country where you've moved to. So you would at least be paying tax on cash receipts rather than on a dry tax charge, which is the problem with the exit charge. Again, like also complicated and probably unnecessary, but just it wasn't actually a terrible suggestion if you had to have some kind of halfway house. Right. Let's move on. Very quickly. Yeah. nailed coffin shut. The AI report from McKinsey, it's a semi nothing burger, but Lomax, there were a few points in here that I thought were quite, I mean, nothing that we don't know really, but I thought it was quite an interesting report.

8:12If you haven't seen it, it was the quantum black AI report via McKinsey. And it's basically saying that the enterprise is really seriously working with AI mainly on the agentic side, which makes total sense, mainly in workflow. Again, makes sense. But it's not working. So Lomax, anything in there on the quick news side? No, I think just quick news is like basically saying almost every large organization says it is using AI now, not unsurprisingly, but only a small fraction say they've generated a meaningful bottom line impact. For example, only 1 % of companies say they've reached mature AI deployment.

8:46So in a way, if you're being optimistic, you're saying, well, there's still a hell of a long way to go. And when we talk about probably later in the show, the huge amounts of CapEx and how's that all going to get funded? Well, then if we're only at 1 % penetration, then clearly there's a long, long way to go in terms of implementing AI and enterprise. So that's probably a very, very good thing. But we are only, but progress is slow, it's for sure. So that's probably the takeaway from that report. The other quick news section I wanted to go into, I first came across this a couple of weeks ago was the the next period um debacle so they're the dutch chip maker that are being throttled so they i think they design in the netherlands and then produce or then they ship to china to be finished produce and then the ship back and they make the mid-tech chips not the high-end chips but the mid-tech chips that like i don't know do the window mechanism in your car or the handbrake mechanism in your car but massively massively important and there's been a bit of political to-ing and fro-ing and other to-ing and fro-ing.

9:51Mads, what's the deal with Nexperia very quickly? So Nexperia is part of sort of the silverware of the European chip industry, semiconductor industry. It's a legacy from, I think, from the Philips days, right, where ASML also came out of. And Nexperia actually produces semiconductors, but they don't package them. and it was sold to a Chinese company. And what happens today is that kind of the wafers are created in Europe and then they're shipped to Asia, primarily China, for packaging and finishing before they're then shipped back to Europe for insertion into cars. And Xperia's got a huge market share.

10:34I mean, that business has got something like 40 % market share in some of these very high volume, low value chips that go into all kinds of automotive aspects. It's kind of the power diodes, it's the MOSFETs for the EV battery management, it's the traction inverters. It's kind of lots of good stuff that helps regulate power in an EV. It's really, really important. Each one of them is not very valuable, but you can't make a car without them. And so the thing is that the automotive industry has got four to six weeks of inventory. And so if supply stops, You can kind of keep running the assembly line for a month and a half, and then that's it.

11:15You cannot make more cars if you don't get more of these things. There's a huge debacle and sort of power struggle between the Nexperia and the Netherlands and China. And at some point, the Dutch said, listen, that's it. We're going to seize the asset. We're basically going to take it over because we don't want to be beholden to the sort of threat of Chinese supply chain stopping our automotive industry. and the Chinese then promptly turned around and said, well, that's fine. It just so happens that every time you make one of these wafers, they still come to China for finishing, and we're going to stop shipping any of them back.

11:51So quite quickly what we figured out was, look, we may have part of the supply chain here, but the rest is in China. And so because we don't control that, we are totally at the mercy of supplies coming the other way. And there's a huge lesson here just about how vulnerable the European automotive supply chain is. And I think this is leading to lots of head scratching around the continent. Well, it wouldn't be so vulnerable if you hadn't sold a critical supply chain component to the Chinese. Yeah, so the truth is that even if you hadn't sold an Xperia, the packaging is still happening in Asia. And this is, you know, two decades of buildup of very, you know, CapEx intensive, laborious, low margin stuff that we were happy to offshore because who wants to do this?

12:42You know, these chips that are selling for, you know, 50 pence or less, right? Super, super cheap stuff. But it just goes to show how vulnerable you then become. And this leans into the whole Chinese vendor program issue. with Huawei and ZTE and others, there are some questions and challenges around pulling these out of critical infrastructure. So Lomax, what's happening here very quickly? So this is the 5G news, right? With Huawei and ZTE Corporation. Yeah, I think it's starting with telco, but it feels like it's a slightly broader topic. In the context of 5G, I think this is now the EU moving into law, what has really been happening in practice in the last five years anyway.

13:27So it's a formalization of what has already been in place. And this is potentially, so this is ripping out Chinese suppliers from the 5G infrastructure being rolled out. Which has been in the news for a long time. That's why I said, I mean, basically, five years ago, we kind of agreed to start implementing this. But now it's actually been giving legal footing. So it will be mandated that the supply chains are free of these vendors. it's potentially a massive headache for telcos in the sense that the ship has slightly sailed in some respects and so I think there's this and other things we talk about today where really over the last 20-25 years we've become more and more reliant on Chinese supplies across everything for various reasons and now we're paying the consequences for that and now we're saying we don't want we want sovereignty over our own supply chain we effectively want you know walled gardens in this respect um but we're making that decision on in the context of 5g when we'd already decided to use these vendors in the first place so i think there are um telcos now in europe who are going to be facing big um big headaches to try and clean up their supply chains.

14:50The Germans are facing a wall of hurt here, close to 3 billion euros, Deutsche Telekom alone, 1.2 billion euros just to swap out some of these switches. So is this trade or security or both? Security. Yes is the answer. Okay, well, we are going to talk a little bit more about tech sovereignty and the like later. And I don't know, Lomax, if you want to just very quickly talk about the Einride SPAC. Yes, aren't we so back now? Spackety back. Exactly. So we are now SPACing autonomous EV trucks. It has a nice ring to it and echo of the 2021 days. So Einride, a Swedish autonomous trucking company founded in 2016, has announced a SPAC in the US at a 1.8 billion valuation.

15:45They will also be raising$100 million at the same time. They apparently have contracted ARR of$65 million. They have trucks in the wild in Belgium and Sweden. They have real paying customers. And unlike, you may recall, Nikola, which SPAC'd in 2020. They're dead, right? They're nixed, right? Yeah. And the valuation of that SPAC was$30 billion. so 15 times more than 15 times the the value of einride um and so that that company ended up going bust earlier this year so we have higher hopes for einride it's a european company so great great job there it raised around 500 million to date so it's not been cheap to get to where they've got to um but actually a nice uh nice story in the news this week that should have been deal of the week my man but um no we've got better deals oh we got better deals exactly exciting um and very quickly uh alex cart palantir you made me watch that video is a sorcery i can't i can't i can't i can't quite watch him i don't know what it is but all i hear is word salad every time he opens his mouth i can't i can't quite tune in i mean palantir you know palantir has been the darling of of stock market even a heavily shorted stock of stock markets.

17:08Yeah, and Burry's on, NVIDIA and Palantir on Burry. Well, the PE on Palantir is off the charts. It's more closer, it has been closer to 200 at one point. Palantir is a hugely influential company that talking of critical supply chains is now selling into, you know, many, many European governments, right? Including the UK government. Absolutely. Cross-defense, cross-health, like two absolutely highly politicized areas as well. um i think that so there's a there's a um book that's come out this week about alex carp the it was his biography right yeah biography yeah yeah and there's also he's done a sort of 25 minute video which is on youtube with molly o'shea sort of journalist um i would recommend watching that for a um i struggled with it man i struggled with it it's for it's for i think it's 45 minutes but i i the guy just talks in waffle to me i i maybe you know you need to certain people can tune into certain people and he's not my guy but all i heard was word salad it's like quite scary word salad to me but you obviously found that's the guy building the critical infrastructure for you know the nation's nation's defense and health so and so i i you know i think I think Palantir has attracted some, you know, I know some of the UK teams, some fantastic talent.

18:30Yeah, it must be. Yeah, absolutely. And it's got, there's a certain vibe and which, and he is the sort of, you know, modern day, you know, I don't know what, whatever word, philosopher King gets thrown around a lot in, in Silicon Valley, but he's certainly like a crazy scientist to me. Or, or a Pied Piper type person, but certainly he galvanizes talent and, and interest and energy. So it's, normally listed companies like this can sort of outgrow those kind of personalities sometimes. And at some point, you need to get somebody a little bit more corporate. But, you know, Palantir just keeps growing and innovating.

19:06And you need someone like that. Yeah, exactly. Yeah, fair enough. He's just hitting his stride. I mean, I'd say it can sometimes, as Dan says, be a little difficult to discern exactly what he means. When I've heard him speak, my takeaway is that he was quite visionary around what it takes to build data-driven architectures. In an age of AI, which is a lot around data, and using data as government would do to understand people, to understand patients, to understand the tax base, to understand intelligence, few things are more important than privacy, and having a system that's designed so only the people that are supposed to see the data can see it.

19:45And what he will say is, nobody comes close to designing and having designed their whole data architecture for privacy like Palantir. So it's the only system you literally can put all this super sensitive data on and feel relatively confident that only the people that are supposed to access the data can access the data. Now, whether that's true or not, okay, it may be an argument, but I think he makes a strong case of the importance of it. And it's clear that he has understood that that is what's important, I think way before a lot of other people did, which is why you can win so much government business.

20:23Let me give you a quote from Karp. So this is why I'm also, I feel very, very aligned with him in some respects. So the modern incarnation of Silicon Valley has strayed significantly from this tradition of collaboration with the US government, focusing instead on the consumer market. A generation of founders cloaked themselves in the rhetoric of lofty and ambitious purpose, but often raised enormous amounts of capital merely to build photo sharing apps and chat interfaces for the modern consumer. Amen to that, quite frankly. So I think, you know, he's a guy building serious stuff that we in societies need.

20:55Okay, well, we are going to talk about him. Not your cup of tea. Okay, move on. Really not. But then I'm so drawn to, I find that there are these incredible entrepreneurial characters, the Zuckerbergs, the bases, the masks, the tears, the him ever. but as soon as they're in a position of power i start to recoil a bit because i can't disentangle as an emotional creature the person and the business and i don't think it's always a useful mindset it's a failing of mine but when i'm i'm hearing engaging with these kinds of characters and they feel like me they're living on planet zanusi somewhere i just want somebody with a sense of duty and care for other human beings to be in charge you get too emotional dan but i think i i mean i don't disagree with you one thing i would say by the way about carp is you watch this stuff and you you know look read it he um he has you know this concept of backing founders with a big chip on their shoulder you know that drives them keeps them up at night uh carp definitely has a few of those in terms of his backstory he's baked he's baked in trauma and you can see it you can see it in his body language in his voice in his eyes there's a there's a fizz there's a fizz to the man as he bounces off the walls.

22:05But it just doesn't sit well with me. And I think there is a layer in there that almost can become musky in West. It kind of pops off the charts on the other side of this success and this power and this money. But we'll see. I mean, I'm going to reserve. He's also very, very proud that he's basically made a whole cohort of investors, what he would say kind of venture style returns in the public markets, which, you know, like he's obviously having a very public battle with the short traders, short sellers at the moment. Which we are going to cover. And thank you very much for the segue because we are going to...

22:41One of our deep dives is the market meltdown. Now, at time of press, it is Friday the 14th in the UK, 4.29pm. And we have seen over the last day, especially 24 hours and for a short period of time, a bit of a market meltdown, some version of correction. I think the only buck in the trend is Apple last time I looked, but it looks like everything else Mag7-esque and tech and AI has taken a hit. NVIDIA being the biggest hit, I believe, at about 10 % down in the last two weeks. So a wobble. Is it a bubble popping? I'm not going to ask either of you to crystal ball, but Mads, can you start this one off for us?

23:25Well, I might crystal ball a little bit anyway. Do it. Love your crystal balls. This is probably just a bit of risk off and people taking some profit after a great year. I think this is what's happening. But there are always stories. There are always narratives that you can use to justify what's happening. And I think there are two vectors that are worth considering right now. One is the one we've been talking about all year, which is Chinese open source models that are incredibly good. You know, we've been saying for a long time that the Frontier Labs in the U.S., they are threatened by the open source models because they've become so good.

24:00A quarter or two ago, they were six months behind. But we've seen Kimi K2 come out in the last week or two being on some parameters better than the Frontier Labs models. And it's an open source model and it's free. So what do you do if you built your whole kind of investment thesis on putting hundreds of billions of dollars behind these Frontier Labs so they could spend money with the hyperscalers, that could spend money with NVIDIA, that they can recycle it all back round. And somebody now comes along and says, well, I can give that tech away, that Anthropic, and openly I have spent so much money developing, I can give that away for free.

24:37Okay, so that is kind of one specter there that's haunting markets. Not touched by tariffs, not touched by export controls. You can go to Hockingface and you can download it. You can run it on your own laptop if you have enough memory, right? You can buy one of those beautiful NVIDIA boxes and sort of a local GPU, and you can do real damage. So that's on one side. It's getting people a little spooked. On the other side, you've got the hyperscalers. And this is where there's been some exciting kind of news these last few days here, because we've got Michael Burry, who was one of the protagonists of the big short, kind of one of the few people that saw the collapse at the great financial crisis and saw these subprime mortgages and shorted them in time, he has said, look, there is just so much accounting fiction that's going on here with the hyperscalers that they're overvalued.

25:31And so what does he mean by that? Well, it turns out that when you buy an asset like a GPU, you don't expense the full cost of that asset all in one go. You have an assumption around for how long you can use that GPU you bought from NVIDIA. and then you're trying to use that as you calculate your P &L. So if you buy that for$100 ,000, you might say, well, I can use it for three years and therefore I'm going to expense a third to say$33 ,000 every year into my P &L and that's my cost. And when I calculate my profit of my business, that's how much cost I'm going to put into that calculation. And over the last few years, as the innovation cycle of tech has gone up and as the release of new hardware has gone up, the hyperscalers have all said, well, we think that the hardware is not going to have a shorter lifespan.

26:27It will have a longer lifespan. And that, of course, is the same as almost making magic. Because instead of saying that these assets will last for four years, they are now going to last for six years. That means instead of expensing 25 % of them each year in my P &L, I'm now only going to expense one-sixth of them in my P &L every year. And Michael Burry, he's looked at that and he's saying, that just does not make sense. It doesn't make sense for a number of reasons. First of all, you're expensing them in a straight line. That means you're expensing the same amount of money every year. And just for easy math, let's say a GPU costs$120 ,000 over six years, that would be$20 ,000 a year.

27:09Well, if the cost of it is$20 ,000 this year, you're sort of saying, that's how much I'm going to expense it against my P &L. There's no way it's going to have the same value six years from now because the chips, the GPUs keep getting better and better, right? There's just no way it's going to be as valuable. So that's the first thing. But the second thing is the gating factor today isn't necessarily GPUs, it's power and data centers. And so what we've seen companies do is they start to rip out the equipment before it has melted, right? They start to rip it out to put new equipment in that's faster because the new GPUs are so much more powerful than the old GPUs that financially you can't justify keeping them for too long.

27:49So at the same time, as all this stuff is happening, Google and Microsoft have both said, well, we think we can use them for six years and we should depreciate them that way. And so Bury and others have looked at it and said, guys, you're overstating your profits. Now, what might really be happening here? So it's possible that everybody's thinking, we just need to show numbers that are good enough for long enough for all the ROI and all the enterprise spend on AI to come good, that this becomes a self-fulfilling prophecy. And if it does, nobody's going to care five years from now that we have to write off these GPUs because we will have arrived at some new economic paradigm where all will be forgiven.

28:34The risk, of course, is that it takes longer and suddenly you're going to be left with a big charge, right? Let's say that you had assumed that you'd expense them over six years, but really after four years, you have to write them off. Well, suddenly you're going to have to take a big hit in your P &L. That'll depress profitability, and the hyperscaler numbers won't look nearly as good as people thought. So this is another thing that's spooking markets. So we've got that on one hand. We've got the Chinese open source on the other hand. And so just enough of a specter sort of, you know, walking around, you know, whistling around here that people are getting a little spooked.

29:08But Barry is great at dropping a bomb and then deleting his Twitter account, isn't he? He's done this a number of times, old Cassandra, where he goes, mic drop, boom, deletes the account. A hundred percent. Look, like many clever people, he has probably forecasted 10 of the last two recessions, right? Yeah, yeah. He did the 20, he did obviously the GFC, although he nearly got in a lot of trouble over the big short, the GFC predictions where he really had to pull out the legal powers to stop people from trying to close down his fund. Was it Scion, I think, which he's just now closing? In investments, being too early is the same as being wrong.

29:47And look, I think this is more of a little hiatus and a little risk-off period than actually a full-blown meltdown. And so it could be that he would have had to hold those positions for another two, three, four, five quarters. And it's just nobody can afford that, right? The markets can stay irrational longer than you can stay solvent, is the old saying. Now, Lomax, before I bring you in, I want to just, I was Googling Michael Burry and just to understand a little bit more about the man. I didn't realize he's actually a medical doctor. He lost his left eye at two to cancer. He's a super introvert, just almost reclusive.

30:25And as we already mentioned, he has put calls on Palantir and NVIDIA. So he is assuming a call off in AI. But Lomax, what else would you dig into on this topic? Well, I think this ties into actually. So we talked about the McKinsey report earlier, basically saying two things, one, which impacts this, I think. One is that there's still a long way to go in terms of AI adoption in the enterprise. So that's good news because it shows that there's still a lot of AI revenue on the table out there. But do you think there's a fear? Very quickly, do you think there is a chance that all these enterprise pilots are going to cause a break in that?

31:05OK, we tried it. It doesn't work. That's it. We're not doing that again. or do you think this is just an iterative force where they're trying, they're failing, they're not implementing, but they're going to keep going? Which way would you go? Well, I've said it here before, is that I do think that there will be, people won't see the productivity gains that they wanted from AI. And so there will be a cooling off at some point, for sure. I think long-term, I'm bullish on this, but I do fear that there's an exaggeration now of the benefits from AI, which may not be actually realized in the enterprise.