This Week in European Tech: Europe has the talent. Can it own the upside?

0:00But I use it as evidence that the talent was never the problem. We clearly had the talent here. Europe produced the people who built Stripe. We just didn't give them a reason to stay. So we have to do the Chinese play, which is use the foreign knowledge, the foreign capital to feed our own ecosystem and then slowly transition and draw out the pension funds to your point. You know, it's not a lack of a legal entity that's the problem in Europe and preventing us building global champions. It's the capital. So you could argue that... The UK's whole approach was on a goodwill basis. And this is the first time that they decided to say no.

0:38And I think, what does this mean? The best model is now gated by nationality. If you're building in Europe, that's a supply chain risk that you can't price in. I just feel that this is an inevitability. And the longer we wait, the more dangerous it becomes for us from an economic growth point of view or from a security point of view. You know, sovereignty in AI isn't just about owning data centers and GPUs. If you control the model, someone else can ultimately decide whether you get to use them. Hello and welcome to Upside, where every week we dig behind the headlines that are going to affect European venture.

1:12This week it is pre-Andrew and my good self. And what are we talking about? Well, we've got the future of innovation in the UK. According to Anthropic, what will our economic future look like? AI doomerism loads in the press this week. Both sides pile in, but there's a twist. What's happening in the week ahead? Predictions plus deal of the week.

1:45This show is not investment advice and the hosts of this episode may be invested in the funds and companies featured. Hello lovely people. Pri, how are you my dear this fair Friday the 11th of September? Hello, I am doing great. Excited to be here. Lovely, lovely to have you. Andrew, you're in Blighty. Welcome my good man with family doing shenanigans before you head back off to Portugal. How are you this week? Very well, very well. Yesterday was cloudy wasn't it? So it was good to come back to the UK and experience what the weather's going to be like for all of you for the rest of the year. while you go back to paradise lovely thank you so much if you can send some before we kick off did you did you see the um i was super disturbed by the the nathan fielder and lance oppenheim's documentary with elizabeth holmes did you see the trailer for the new movie about elizabeth holmes and theranos it was recorded just before she went into jail all secret squirrel prude did Do you see that?

2:44I didn't. I need to watch that. I couldn't work out whether it was AI or whether it was some kind of parody or whether it was some kind of outer context clip. But if you haven't seen it, it was a movie launch at Telluride Film Festival. Looks super interesting, but it makes me feel a bit sticky. What did you think, Andrew? Art imitating life imitating art. And the other one that came... Yeah, it was a bit freaky. The other one that came out this week, or the other trailer that came out was Artificial with Andrew Garfield. Did you guys see the trailer for that one? Which is the Sam Altman story of him being fired and rehired in OpenAI in 2003.

3:25That I think is always funny. Yeah, I don't know when the movies are out. I think Artificial is out on Christmas Day or some super random day like that. But there's lots of stories to be told. It's funny seeing Jesse Eisenberg from the social network not want to meet Mark Zuckerberg. And then Andrew Garfield talking about not wanting to use chat GPT. Obviously, very creative souls, very emotional souls. But it was, I don't know if there's a story there. The Apple event happened. I am super fizzy. I can get my foldable this year, October the 23rd, I think. I didn't see anything else that I think is worth talking about on the Apple event.

4:00But either of you found anything in there? On Apple, we've got the personal AI assistants. we've got well apple muse and instinct that we spoke about last yeah i downloaded instinct this week but i didn't put my credit card in i i got the fear i started sending me insights into my inbox and messages and then i got the fear and didn't put my credit card in well funnily enough that you say that um we're talking about apple and i think apple are a bit behind on this on this sort of personal ai agent race but muse has built and metas muse which launched on the 8th of September, that in terms of payments, it makes purchases through Stripe.

4:40And there's like a layer of protection there so that customers feel safe. So they've already anticipated that customers will not feel safe. So that makes it a much more secure shopping experience. So interesting, you said that about instinct and not putting your car details on. I think we're just becoming much more attuned. Or maybe this is just because we're on the inside track of tech and VC and investing and whatever. But I feel it feels like people in general are much more attuned to trust. So I do think these releases are not coincidental with Apple releasing iOS 27 and obviously their big event because I think they are going to own this ecosystem, this personal assistant ecosystem.

5:21So I can see why Muse and Instinct are coming out in that same week. But I do think people are going to be so, so connected to trust. And there's been so many kind of whatever open claw issues. So we'll see how those things are managed. Maybe it's just a simple case of gatekeeping every transaction. So don't buy the airline ticket until I've signed it off. Maybe it's that simple. A few other quick things, market and pulse check-in before we move into the main docket. So ECB rates went up 25 bps, now 2.5%. Inflation is still sticky across Europe, which means that prices, mortgage rates, interest rates are going to go up.

5:57US Treasuries are close to 5 % this week on the 10-year bond. Brent's now over 100 bucks, which is another inflationary push. All bad news for us, therefore all bad news for founders. The US and euro markets slid a point or two at time a pod. Nothing major, but there's a little bit of a slide, a little bit of a neutral to slide. The Oracle results came out. The Adobe results came out. Adobe, I think, was a bit of a nothing burger. It's our kind of our sassy check-in as what's happening in old world land. But Oracle is a bit of a canary and they did well. So total revenue is up 30 % to nearly 20 bill, three months up to August, exceeding analysts, and then shares jumped around 6 % at time of pod.

6:39Now they've got a backlog of about$664 billion in contracts, which I think is being weighted here, but we will see how much of that materializes. The other thing I wanted to just quickly highlight at the same time, some reports coming out about European tech and how much they raised in August. So European tech 3.2 bill across 165 funding rounds in August, which is down around eight from July, which took 267 deals. Now, this is a very common seasonal dip. All of the investors do head off to their yachts in Central Bay. So there is obviously a European dip in August and everything normally bounces.

7:17This is very common, happens every year. Amazon, they are raising a 4.2 billion, their first pounds sterling bonds. So they're doing any currency they can basically. So this is the pound deal. So they're going all over the world raising. This is a, what is this, 4.2 bill. It's roughly four bond buckets across four year structures. So there's a three, a six, a 12 and a 19 year bond structure, roughly a billion in each bucket. The 19 year bond is, has a 7 % or near 7 % yield that they reckon. That's quite a chunk of change. Did you see, boys and girls that Revolut won their conditional US banking license.

7:56Did you see that? Yes. That's kind of cool. HR tech, one for you, Andrew. So Jack and Jill raised 30 mil and HR tech, and this kind of spiked you a bit. What was your thinking around HR tech? It's an enormous industry, but so many people are trying to crack it fundamentally. And so, you know, we've seen the digitization of HR, but I just question whether what is a very old school person-to-person industry can be when you can remove the person from the chain, especially when you look at sort of headhunting versus marketplace allocation. So obviously we've seen lots of platforms where, you know, you go and post jobs, where you've got a slightly different business model, where, you know, on demand and the task rabbit type up work, the places where you can go and hire a gun for hire.

8:45the hr industry is extremely old-fashioned in the way it works they use digital tools internally for their for their process and flow and management and data but actually a lot of it is still taking people out to drinks on on friday friday lunchtime um taking up you know keeping keeping the customers happy the tech head is saying don't use tech is that where is that is that where we're going with this i don't think it's don't use tech i it's not a technical challenge anymore right it's a question of whether the human beings want to change behavior and just use technology to hire even if it is driven by ai rather than actually talk to someone about it and rather be taken rather than be taken down the public Thursday afternoon I think the um what the atomico state the state of tech report is going to be out soon kind of September October and they led the round in Jack and Jill I think so maybe we'll see maybe we'll see a little bit of insight because I'm sure they'll be singing the bullish praises of HR tech in there maybe.

9:46So we'll see. I think they're great tools there. And there's another one in this space called Hire Stitch that I've been tracking. And I think they did their raise a few months ago. They raised$25 million from A16Z who backed them. So it's starting to be a competitive space, as you say. There's quite a few players out here in this space building HR tech. But I think it's great for skill matching for employees. and I think Andrew you say you know it's it's all about taking someone out for a coffee or to the pub on a Friday but in terms of skill matching I think it's phenomenal we're taking out that sort of highly labor intensive part of the recruitment process by getting AI to skill match and it is very capable of doing that I think it's a it's a great solution at least for the initial stages of screening and maybe less so for the interview stages well provided the person hasn't lied on their cv of course that's what the interviews are for well this is this this is the big challenge isn't it with hr a lot of it is culture and it's one of those things that you can't of course we look for skill and aptitude but we also look for drive ambition culture which you maybe can see some proxies for but they're really hard to lean into digitally so you're always gonna you'll always need that human angle like you say andrew and also i think there's there's just there's just not the volume of cash going into these spaces.

11:05So they seem to be almost like the middle of the barbell hedge at the moment, rather than the driving force, the apex of what people really want to invest in. The other thing I saw this week was the very unloved article titled 100 Days to Save EU Inc., which is a bit click-ragy. So this EU Inc. is the landmark law for standardized Europe-wide company registration and taxation push by the EU Inc. guys, which then the name of the legislation became EU Inc., which is kind of cool. And obviously, the fear is that the longer it stays in negotiation, the more watered down it becomes. And the theory on the title was that this has to really get into law by the end of the year before Parliament goes into snooze mode for Christmas.

11:53So I don't know if anybody has any thoughts on this, but I wanted to raise it because I think it's important that it happens. European fragmentation is such a massive challenge for us as a European, and I'm including the UK in that, as an industry. So we need this as law, not as kind of guidance. But I don't know if anyone has any thoughts on this one before we move on. I guess there are asks in the EU Inc letter that went out that said, and I think two of the signatories that pushed hardest on this were about EU-wide registry instead of a business having to register in all 27 member states. And then another one was stock options that are taxed when you sell rather than when you're granted them.

12:33So some EU countries will tax when you're granted shares. That was another another key signatory and the proposal is let's let's allow you to do this incorporate a business digitally in 48 hours like in Delaware in the U.S. and I guess that the point is the U.S. already has this and a really really good example of it is Stripe Atlas founded by two Irish brothers and incidentally sell a product that allows you to incorporate a Delaware company from anywhere in the world in a few hundred dollars in a few days so the thing that the EU Inc. is trying to legislate into existence. America already has done this, and it's so standard that it's being sold as a product off the shelf.

13:08This is, I think, the frustrating thing for Andreas and the guys is that there's the template. We know how to do this. This is how to do it. First of all, Andreas, I mean, he's done an awesome job driving this, so hat off to him taking on the EU and trying to do this, and it's a fantastic vision. But, you know, my understanding last time I deep-dived on this, that the fundamental weaknesses in article four which i think subjects all matters not covered by the articles of association for the eu inc to the national law of the of the state so your challenge becomes that anything that this this legislation doesn't cover you just flip back to dealing locally with the with the national government i think that's where the sort of the the analogy with with delaware somewhat falters the devil will be in the detail i'd love to see it worked well, but I just doubt whether the red tape, which is Europe, and all the differing handling of tax and shares within the member states will mean that this is something that people jump on.

14:13Because you could end up, ironically, you could end up with a more complex situation and certainly unknown in the short term than you are if you just pick a UK limited co or something that's a known quantity. I wonder if there is a midpoint. Now, this is a random thought bear with me but i wonder if there's a midpoint where if you just get some version where the member states can apply the best most useful version france germany spain italy whatever then that will draw up the rest or maybe even less sizable states and that will draw up the rest to behave like others and therefore people can effectively become whatever whatever founding company in whatever state is doing the best version of it so i'm wondering if there's like a if not everybody's on the same level maybe there's a midway point where five six seven eight nine member states kind of do a sub club i know that there was some talk of something similar to that a few months back but i don't know if andrew maybe you've got some thoughts on that i think the main thing is you've got to start somewhere right so it's all it's all very well saying it's not perfect but if this is if if it does become popular then it's it's a bridge it's a bridgehead to somewhere else um yeah the the the counter view is that this isn't the problem in europe you know it's not a lack of a legal entity that's the problem in europe and preventing us building global champions it's the it's the capital so you could argue that you'd be far better off championing european pension reform and getting that you know at 0.01 up to 2 % like it is in the US and focus on that rather than worrying about the structure of limited companies.

15:58No, I think it all comes down to money. So I don't think anyone's going to argue with that. I'm hoping that both beasts have been awoken and that it's not going to be another decade. There was a report, Yoram at Deal Room, they launched a new report and Mads isn't here. So he can't punch me on the nose for talking about reports because he hates reports. but Pri the the the essence of this base is basically that two trillion dollars worth of unicorns have been built by European founders abroad so it's just a window on what we've just been talking about but where would you take this one yeah I'd say a couple of things on this I think everyone's using this chart of I guess Labour advances lost as a list of of what we've lost but I use it as evidence that the talent was never the problem we clearly had the talent here Europe produced the people who built Stripe.

16:47We just didn't give them a reason to stay. And of course, we talked about some of the issues and the reasons behind that with the EU Inclus. So there are clearly reasons why talented founders are leaving and going off to the US. Sadly, yes, we're losing talent and companies to the US. On whether it will always be the case, and we're always going to be losing founders and businesses to the US, I think that there are some companies that are still proving that Europe can win. For example, on the positive of note, Mistra raised 3 billion 21 on Monday, which is the biggest round in European tech history here.

17:20And ISA got to orbit on Saturday. And I guess the part that hasn't changed is the paperwork. And that's something in front of Parliament right now with EU Inc. So that like we have got regulatory or legislative challenges here. The other important thing to note is that, okay, so we do have some businesses that are successful here that have been built in in Europe and are staying in Europe, but the capital is actually still coming from abroad. So when you look at the biggest round in European tech history being Mistral was co-led by Samsung with A16Z, NVIDIA and BlackRock in it. N-Scales, 2 billion is largely NVIDIA.

17:56And European businesses are being capitalized, just not obviously by Europe. The checks are coming from abroad. So we have to do the Chinese play, which is use the foreign knowledge, the foreign capital to feed our own ecosystem and then slowly transition and draw out the pension funds to your points. But you mentioned ISAR, and we've got to talk about that, Andrew. Space, your happy place. SpaceX, European rival, or is it? Tell us what happened this week with ISAR. Well, it's the first ever European launch into orbit, where we actually launched something that's actually got to space commercially, and actually ever from the European geography, because all of the traditional European Space Agency launches have come from French Guyana.

18:43You know, Britain, who's the only country ever to give up an indigenous launch capability in the 70s, used to launch from Australia. So, yeah, first time a rocket's actually got to launch something into space from Europe, which is very exciting. I think it's worth stepping back and thinking, you know, why is space important? And space is becoming a trillion dollar industry this decade. It's growing roughly twice as fast as the wider economy. We talk about space a lot and you think about people going to the moon. But what else is space useful for? And it's actually a huge list of things that are evolving and will evolve in the space economy.

19:16And I'm just going to rattle some off, like, you know, communications, that's broadband, mobile, secure networks, Earth observation, imagery, you know, weather, climate, navigation, defense and intelligence. So sensing communications, you know, now the possibilities of data centers, you've got manufacturing where you can do things in zero gravity that you can't on Earth for new types of drugs you can deliver differently or semiconductors, potential energy for solar power, space infrastructure. So space stations, habitats, you've got tourism, you've got extraction of minerals in the future from either from Luna, from the moon or from asteroids.

19:53so you know and then you've got launch itself which is obviously a commercial industry so space is super exciting and ultimately from a sort of societal point of view obviously if you control space you control planet earth which is quite sobering when you think about it so that's why having access to space not just commercially but as a nation is really important and if we quickly sort of run through the status quo you know launches last year i think about 190 for the US, 90 from China, seven by Europe, and about 25 elsewhere, India and others. So about 320 odd launches last year, we'll punch through that in 2026.

20:31But when you look at actual mass, so the amount of stuff, the weight of stuff launched is about eight times as much by the US than anybody else. And that's because, you know, they've got three launch systems that can launch, you know, more than 40 tonnes into orbit. And that's important for when you think about, you know, crude launches, when you think about lunar modules, large military satellites, that sort of stuff, and constellations of satellites. You know, if you can't launch, you know, ISAR, which is what started this topic, you know, is a one-ton rocket. Starship is 100 tonnes. So you can look at the disparity there.

21:08And I think it's not that people, you know, a one-ton rocket isn't useful. it is and often for governments for small smaller satellites governments or commercial operators will pay a premium not to wait for a ride share and you know that's what arianne and and um things like starship are you know that they're if they're not large constellations it's not one big payload it's a lot of customers on one ride into space but equally you know small launch economics remain pretty brutal and most satellites can fly much more cheaply as passengers on you you know, a Falcon 9, Falcon 9 Heavy or an Ariane.

21:44But they're not taking any traffic, are they? I mean, I'm pretty sure that SpaceX aren't taking any other cargo apart from their own for the next maybe 18 months, two years. Well, no, they're taking other people's cargo. But I mean, there's a weight, right? People can't get stuff into space. I mean, for two years, it's pretty full. So if you want to get into space, it's really tough. And that's where the opportunity is. And I just think, you know, the analogy is that container ships, you know, when they came along, didn't just carry sort of existing cargo more cheaply the ability to move mass super cheaply changed what it what made sense to trade and how trade worked and i think that's what's very interesting about the thesis yeah yeah exactly so look this is a fantastic first step forward but nobody's really got their eye on medium heavy lift you know starship's already or spacex rather is already launching every other day they'll soon be launching every day even with this ice rocket, I think they're talking about sort of their goal is 40 launches a year, but it's one ton, you know, and they may do a heavier rocket, but they're not going to be building a rocket that can carry, you know, 20 or 40 tons.

22:50And that's what we need to unlock in Europe. And I would argue even we need to unlock as the UK, there's an opportunity, but there's no nobody, nobody batting for that yet. But the tech capability is out of the bag. China is now launching very similar looking rockets to SpaceX. So obviously, technically, this isn't going to be another decade, 15 year plot. We can accelerate the journey to build the higher, whatever it is, yield rockets that can take more stuff to space. Is that fair? Yeah, I mean, as I said, the U.S. has sort of three systems that can launch more than 40 tons or around 40 tons.

23:29China, the heaviest rocket can launch about 25 tons. Ariane launches about 20 tons. and ISR Notch is one. So yes, the technology is there as ever. It's about desire and capital. Will it be driven by defence is my assumption. This will be driven by a sovereign defence resilience push. That will be the driving force would be my assumption on this. One last final comment on the kind of the quick news of the week was there was also a report this week that it's been calculated that the world's governments have two trillion in debt. I think we crossed that threshold recently and somebody pulled together all the figures.

24:07So total global public debt now exceeds$100 trillion, which is roughly within shooting distance of the total global GDP. I don't know what that means, but it sounds a little bit scary. As I mentioned last week, I've been keeping my eye on bond markets and trying to understand the depth and the breadth of how it all fits together. so we are going to be walking into very unknown territories if we're not there already right we are going to have a little bit of an ai corner without mads so first up anthropic one for you so they withheld their latest ai model from the uk testing agency what happened here yes that's right so the uk has a body called um the ai security institute that tests and frontier models before launch voluntarily.

24:58There's no law around it. And Anthropic gave it early access to Mythos, their more powerful, less restricted model. And in July, the AI SI tested it and found it had gone after an open source project, invented fake people and tried to pressure a real maintainer into merging malicious code. And it failed effectively. And then on the 1st of September, which is the one you're talking about, Anthropic shipped the next version, restricted to better US organizations and they didn't let the AI Safety Institute test it and they haven't said why. Yeah, I'm not going to claim those two things are connected, but like the real point is that the AISI never had the right to access.

25:39The UK's whole approach was on a goodwill basis. And this is the first time that they decided to say no. And I think what does this mean? The best model is now gated by nationality. If you're building in Europe, that's a supply chain risk that you can't price in. We also saw, I think we saw the restriction of models from Europe earlier this year in summer by Anthropik. And I think that that is a little bit scary. This is something that Andrew's been talking about for years, this whole sovereign, sovereign control, because we are now, as we've spoken about many times, we're now at this threshold where we're starting to see so many AI attacks, hacks, cyber issues because of the skill and the depth and the breadth of these tools that, of course, countries are now looking to how they can control.

26:31And we are obviously at that threshold now. Andrew, did you have, because I know this is a big thing for you. Did you have any thoughts on this one? Yeah, I mean, it's almost inevitable, I think, that, you know, America will restrict solar models as they become more and more powerful, much as it restricts sort of advanced chips, weapons or nuclear technology. And you're right, Dan, this is something we've talked about a lot. And I just think, again, it doesn't take a genius to extrapolate the line here as AI gets more powerful that this isn't going to be something that is the right of every human being to access.

27:07And that's true either from a commercial cost point of view or from a government deciding that they don't want it in the hands of bad players. So again, I just feel that this is an inevitability. and the longer we wait, the more dangerous it becomes for us from an economic growth point of view or from a security point of view. You know, sovereignty in AI isn't just about owning data centers and GPUs. If you control the model, someone else can ultimately decide whether you get to use them. So, you know, equally, sort of, if you want a contrarian point, you know, you can argue, well, maybe this is just a sort of temporary security policy, you know, about more about sort of US protectionism and Trumpism.

27:48But I have my doubts. I just think this problem isn't going away. And the idea that people will, even commercial companies, will voluntarily kowtow to some safety testing that the UK has decided everyone should do is a bit naive, I think. The only thing that I've been starting to dig into a lot more is the algorithmic efficiency gains. And I was looking into 1-bit LLMs and sub-1-bit LLMs and how much efficiency we can get out of the models and the algorithms as opposed to layering in the hardware. And there are some very interesting, I mean, it's Prism ML, wasn't it, that I think raised at the beginning of this year, who looked at on-device LLMs.

28:33And I think that, did they get bought by Apple? There was some kind of deal, I think, with Apple about how they are going to effectively create frontier level skill and AI models on device that can run on a standard laptop or a standard piece of kit. So I think the other counter to follow on from your point, Andrew, is to, yes, we're going to have this cat and mouse now, but there is going to be, or there will be many more breakthroughs on the efficiency side, where you can get extremely extremely capable models without the crazy latest and greatest from NVIDIA. So I'm going to learn more about that and share more about that as we go, because I think that's a super interesting aspect.

29:13Well, 100%. And a bit of a plug here from one of our portfolio companies, Plumer AI, who've been going a few years, that's a British company, and that's what they specialize in. So it's one bit binary neural networks. They are deployed across edge devices, things like door bells that you can do computer vision processing without having to go back to the cloud you know other devices very very clever technology it results in a sort of 30 times smaller memory footprint it reflects directly back into the amount of energy that you're using so it's inevitable i think that you'll see edge devices and cars and other things leverage the sort of technology you're talking about be it one bit neural nets rather than 16 bit or or other technology and the processing capacity and capabilities will match the feature that you need and the problem you need solving.