E695 | This Week in European Tech with Dan & Mads (feat. Sam Marchant)

0:00Welcome to Upside, where every week we dig behind the headlines that are going to affect European venture. Today, we are talking about Anthropics raise and spend. Is AI making us do more? Alphabet's 100-year bond. We got a bit on European sovereignty and some space, space, space. And we have a special guest today. We don't have Limex Energy. We have Sam Marcher. Big round of applause.

0:36Sam, you are a recent Dubai expat, and I'm claiming you now as our man on the ground in the know who knows everything. We met you many years ago. I think you were Ascension, then went to Hambro, Narsalaka, and now you're wearing many hats. You are building AI startups, investing in founders through Forward Pursuit, hosting the Now We Build podcast. And now today you've given up some of your precious time to be with us. Thank you so much for joining us. How would you introduce yourself and how would you position yourself in our world? I think you've done an amazing job. I definitely can't match that energy for starters, right?

1:14And I don't know, a few, four or five copies in today. There's no way I'm going to get to that level of energy. And look, it's an absolute pleasure to come on and chat to you guys. Often you get asked to go to podcast, okay, we have to talk about this topic, this and that. When you WhatsApp, Dan, I was like, great, I get to sit down with two friends and actually just talk about some fun stuff. So that's quite easy. So yeah, I'm delighted to be on. Thank you so much for joining us. We've got, we're bouncing around a bit today. I mean, we are going to go into some of the AI tech crazy stories. And I think we're going to start there with you, Mads.

1:47And so we're looking at Anthropic. There were three things that really caught my eye this week. So Anthropic, raising and spending, a massive oversubscribed round there. There was an interesting report from the Harvard Business Review on AI. And is it making us do more? Generative AI isn't reducing work. It's intensifying it. So just have a little bit of a chat there. Alphabet's 100-year bond was the other thing that caught my eye this week. it's properly put the whole organization into a whole new orbit we will talk about space laser a whole new realm so having a look at alphabets 100 year bond it's proper you know too big to fail utility now so why don't we start at the top mads give us give us the anthropics rundown what's cooking over there so anthropic it just i mean we've been talking about the company for a long time and they just keep doing better and better.

2:40It was clear in 24 when Sonnet 3.5 came out that they were onto something. And ever since then, they've been really powering ahead. Last year, they were rumored to be working on a$10 billion round that then became a$20 billion round in January and they've just closed it at$30 billion. It seems investors just can't throw enough money at the company because it is so hot. AI is the hottest thing on the planet, and Anthropic is the hottest thing in AI right now. Many, many interesting things here. It's a company that was founded by kind of renegades, defectors from OpenAI, that set up the business just a few years ago, and they've now raised more money than OpenAI.

3:23They've just pipped them. Isn't it funny, Mads, how it feels like everything was OpenAI, obviously, November 22, here when it would chat GPT, it almost feels like Anthropic has stolen the thunder. Is that? I mean, listen, maybe Gemini next year. Listen, it's not the last story in this, right? And Sam certainly hasn't given up and he's going to come out with a round and he's going to raise money. Everything as we've been talking about hinges on whether this move to ads is going to pay off for him. If it does, opening eye will have a new lease of life and that's going to be a transformed company.

3:58If it doesn't, he is in a world of hurt, I think, because Anthropic, they are really heads down executing on this enterprise encoding playbook, and they're doing really well. So super exciting. I mean, we're talking about$14 billion of run rate revenue right now, 10 times annual growth. Claude code right now is rumored to be around$2.5 billion run rate. Enterprise customers are growing like wildfires. of customers spending more than a hundred thousand dollars a year so yeah so much happening across the board sam what um when you move you invest across the board right you do all kinds of investing it's not just ai and but what when you moved to the region were there any i guess was it the same kind of strategy thesis was it all ai centric over there it feels like it is coming looking from this direction but do you get the same kind of stories about anthropic and is it making the news cycle over there yeah it is 100 and i think look the relationship between the uae and open ai is kind of well documented right um you know the region and how they support sam and and what they think about open ai they've kind of made their bet right open ai is the leader here that's where the money has gone if you then look at i mean i posted about this yesterday actually, because I found it super interesting.

5:21Anthropic is part of that round, the QIA, right? So the Qatar Investment Authority have gone in on this$30 billion round that Anthropic have just raised. And I reckon we're going to start to see little kind of wars happening here with the capital pools that we have in the region. When it comes to AI in the region, it's a really interesting question, because what you got to remember is that because of the kind of labor force and the labor structure here, right? And especially with salaries and how they structured things, efficiency savings are not always what companies and executives want to get from AI, right?

5:54The efficiency saving is often not that big of an advantage, especially if you've got a skilled workforce, but actually, you know, from a salary perspective, it might not be the most kind of costly kind of team. So then you say, okay, well, AI has to deliver upside value, right? It has to deliver revenue growth, it has to deliver, you know, a totally new value driver to the business for the adoption to actually work. And I've seen in this region in particular, the AI companies that are doing the best are the ones that bring some sort of outsized value creation rather than an efficiency saving.

6:25And it's really interesting when you consider that because it changes the dynamic about how you position these tools, how these tools raise investment because of the metrics that they're then kind of benchmarked against, and ultimately, you know, how these companies are going to grow. So I think it's an exciting time to see it and And the AI adoption here in the region, and obviously speaking particularly about the UAE, is very, very high. But my question to you guys, actually, on this whole open AI anthropic debate is, are you still using ChatGPT? Or have you gone over to Claude? Or what's your preference?

6:56I was speaking to you. I mean, we don't really matter because we use everything. Mad is a massive, I will speak for himself in a second. And he's a massive Claude fan. And he lives and builds workflows up the yin yang and amazes us on a Monday when we all collect. He goes, oh my God, he's fizzing with excitement about this new thing that does this new thing. But we use everything. But it's really interesting. I was with two founders yesterday who we would speak regularly about using ChatGPT to do stuff. And without us even broaching this topic, we're talking about Claude doing this and Claude doing that.

7:30I'm going, hang on a second, guys. And it's so commoditized. People will just go where the juice is. They'll follow the flow. They'll do whatever is required to get the best output. So that, I think, is a super interesting dynamic that you don't just have Google crushing AltaVista and Ask Jeeves and all the old. It's like, I'll just go where I get the best value, what creates the most magic. And I think everyone, listen, in our world, we're maybe a bit myopic, but everyone is moving over to Claude for sure. are you doing that even though you're in open ai territory have you a claude boy yeah i was waiting for mads to jump in and give us some big flash of insight about why claude is better but um you know he didn't bite uh no i look i i literally this week on the on the home page of my iphone i finally you know you know it's like edit home page i finally removed open ai from the bottom and i was like okay cool claude is coming in open ai is out and um yeah i have to say from an output perspective on just like everything yeah i find it much sharper much clearer it also doesn't just reinforce everything i say which is actually quite nice um you know you you want some of the challenge to come up with yeah exactly exactly mad i don't know if you've got any listen i've maybe i've already primed it that you know this is this is how we roll but i don't know if you had anything to add to the claude story The journey I've been on is that going back to 24, I found first that it wrote better and then it coded better.

9:01And then the models have just progressively gotten better and better. And then in, you know, last year, they released flawed code. And I've really not looked back since then because that's a whole different agentic way of working, right? That's, you know, you set up the skills and you create a lot of the stuff that I think they now have tried to bring to kind of the masses or lay people with flawed co-work. And it's developing and it's coming. But yeah, I think once you're in that cloud code mode where you get the command console and all your skills and all the things you do, there's really no looking back.

9:30It's just incredible. Maybe that's the mode. Maybe that's the capture. Interesting. I want to talk about the HBR report, Mads. So tee this one up for us. It was a, maybe this is no great surprise to people, but effectively saying that AI isn't giving us more time, you know, making us more efficient in a, I don't know what the right turn of phrase is, in a more relaxing way, like we can just do more and our work steady state doesn't increase and everything's fine here. It's making us more intense, helping us do more. So we're doing more, which feels like that's what tech has done ad infinitum going back in time.

10:05But is there something in here that we should pick apart? There's this wonderful book published a few years ago by Daniel Pink called Drive, where he explores the three kind of foundational elements of what gives us drive and life. And it's kind of its mastery and its purpose and its autonomy. Purpose is extremely important, right? And I think everybody who's an entrepreneur, we sort of feel we have our own purpose. We are going out into the world to do stuff because we think it is important. What AI then does for us is it gives us incredible autonomy because suddenly you don't really need anybody anymore.

10:44I mean, you don't need a developer to code up the thing you've dreamt up. You don't need a researcher to go and research stuff. You can just do anything you want and you have autonomy around getting it done and you have mastery of all these skills. And that means in terms of motivating us to do stuff, we have never been more motivated as long as we feel we have a purpose. And so if you're any kind of knowledge worker, whether it's you're doing PhD level research or you're an entrepreneur building a company, we feel so powerful now. I mean, I personally, I have never had more fun in my life. The feedback loops are so short.

11:22It's a constant dopamine hit. Kind of the distance from dreaming of something or having an idea to want to do something and then getting it done is so short. And so, you know, even historically, you might get a little bit tired or you're like, oh, my thinking is not a shock. Well, you know what the AI is really there to give you, you know, a lot of support, even on the thinking and on all the masses of data you got to process. So that means that you can go on for longer. You get this constant dopamine hit. And I think when you look at the HBR study, that's really what it's getting to is that work is more fun than it's ever been before.

11:58And therefore, we end up working a lot. And is that a good or a bad thing? I don't know if we'll figure out. And, you know, 10 years from now, it was terrible. And it made us all into, you know, drooling zombies. I don't know, but I think right now it just feels like what an incredible thing. It's not like we're sitting on TikTok and just passively scrolling. You're actually doing stuff. You're building things. You're making things happen. I think it's very exciting. Sam, what are your main – I mean, you're a man of many hats. You're doing lots of things. You're investing. You're building. You're still relatively new to the region, I imagine, building network and exploring, researching people and things.

12:33but have you got any specific use cases that you are an AI fan of? For me, a big thing, I mean, obviously, when it comes to the massive benefit and upside of AI, I think from a creativity, like as we just spoke, right, whatever you're building, whether that could be something media related, tech related, etc. For me, the biggest use case has been in media, right? And how can we not only call from more information sources, How can I then actually look across this region where news is published in many different languages? News outside of this region, about this region, is published in many different languages.

13:09How can I then not only track that, how can I then kind of take the insights from that and then create it into some sort of content that I find usable? That's obviously the first thing. I am what I would consider to be a very arguably boring, low-level user of AI. And I know that might come as a bit of a surprise, but hear me out here. I like to test AI before I would buy most things, right? So for example, at the end of every day, I will now sit with now Claude, it used to be ChatGPT, and I will just run through everything I've eaten in the day and ask Claude to be my nutritionist, right? And, you know, even here where you could access, you know, a personal nutritionist relatively cheaply or a personal trainer, et cetera, relatively cheaply.

13:53I'm like, okay, let's bring AI in and let it take me 60, 70, maybe even 80 % of the way to figure out whether or not this is actually worth spending money on to get someone who's arguably a professional with this. What I find now is that because of the proliferation of AI, even if you do bring a professional in, you end up kind of getting the same output anyway, because that individual uses AI to just help scale their own operation or what they're doing. So I'm not sure it necessarily adds a huge amount when you do go to, I don't know, like a real human for something like that. So those are the use cases really, really for me.

14:30I like testing my ideas with AI heavily, right? So just, oh, I have this view or I've read this article. What do you think about this? These are how I'm thinking. And just those kind of basic assumptions. But for me, it has impressed me how far you could go with AI before you need to make the kind of real human physical jump. And that, you know, you go into conversations more informed. And I'm sure you guys, I know you guys at Super C kind of go into investment conversations with that prepared mind thesis, right? And it's the same for me. I just find that it's a really useful tool for kind of sharpening yourself up and getting you 60, 70, 80 % of the way.

15:07I think the issue comes when you believe that it can take you 100 % of the way. And I'm not sure we're there yet, right? That's my belief. Let me build on that for a second, because, you know, I think in many ways you're right. There's still definitely, of course, there's a place for, you know, qualified physicians, got a doctor, studied for many years. But have you ever had a situation where you or maybe somebody you knew had some kind of ailment and they'd go to a specialist and they would diagnose them and they would get one kind of diagnosis, one prescription. And they'd go to another specialist and they'd get a slightly different diagnosis and slightly different prescription.

15:42We've always had this idea of getting a kind of a second opinion. There's nothing wrong with that as such. But kind of my impression is that, you know, even the very highly educated specialists, they are debowned by kind of the human condition of only being able to hold so much in our heads and only being able to read so many articles and only having so much knowledge. Now, take the long tail of that. Take all the nutritionists and kind of people that don't have a seven-year medical education, but that are advising you on the biology of your body and how different things functions. Now, some of those people will be very well-read and have a very good idea of stuff.

16:19But my suspicion is, and it's maybe also personal experience, is there's actually a lot of people that aren't academically that rigorous. Now, contrast that with what you can do with an LLM today, doing deep research, getting an autoscan, every single academic paper on a topic, giving it precise input in terms of your blood test, your biomarkers, giving it your DNA. that I had a situation where I had a particular deficiency of a mineral. I fed it my DNA tests, and it quickly came up and said, well, listen, you've got this particular marker there, this variant. That means such and such and such according to these academic papers.

16:56Do you know what? I think if I was going to go through and try and find specialists that could advise me on that, that would have been quite a complicated process. I'm not sure my local GP could have done that, at least not without taking several days of going through stuff. And so, yes, of course, there's a role for qualified physicians. But when we say the 60, 70, 80 % of the way there, sometimes I feel it can take us much further than the type and kind of advice we could have gotten and had access to before. And I think that's really exciting. I agree. I think it's super exciting. I think my skepticism on that like 10, 20, maybe even 30 % less is because of the fact that these human beings were often very happy to make a clear-cut decision, right?

17:41And we'll say, okay, I'm standing behind X, Y, and Z research or experience, and this is what you should do. I'll give you an example. As recently, I was with my dad, and he was looking at a particular stock, and he put it in. He was actually using Copilot, and he put it in, and he said, hey, tell me what you think is going to happen with this stock. And I said, well, obviously, it's going to come back to you, and it's going to say, it could go up, it could go down. And naturally, that's exactly what it did do, right? It didn't really have a strong opinion. It couldn't really make a clear cut decision.

18:09Whereas, you know, had he gone to an IFA or a wealth manager, they probably maybe also would have said the same thing just because of how those guys operate. But at the same time, they would have had a stronger opinion either way and maybe a little bit more bias. I think in the nutritionist, like again, maybe we're going off topic here, but like the physician, clinician, any kind of in anyone in that space, what I like about the tools is that they often don't come in with like a, what I would call like a research bias or research relevancy bias, which is they're not just going to give you the insight that they got in the last piece of research they've read, right?

18:42And therefore they're saying, this is what you should do. And it's not actually what you should do. It's just, it's the last thing that they read or that they consumed that you should do. Does that make sense? And I think, whereas with AI, what we're doing is looking at it and going, okay, I have access to all of this information with a far greater degree of context. And therefore, I'm then able to make a kind of personalized recommendation and tailor that. And that's exciting. That's exciting. I just think it's interesting, you know, and again, we're in a bubble somewhat, right? Because we're using this tech every single day.

19:11And when I speak to friends who aren't in this world, you know, the stuff that they're using, GPT or Claude for, you know, it'd be recipes or, you know, can you tell me the best restaurants in this area? I would argue more search engine rather than thinking about how does this replace or how does this get you part of the way to accessing a knowledge worker and their insight so yeah it's a it's a really interesting time and you you definitely i think from our perspective see a wide range of kind of use cases and and it's really interesting to see i think how people use these tools it's going to be fascinating how i mean using health as an example how you get such Such a broad context.

19:51You get to see the world through not just one vertical specialism, but how everything interconnects. And I think that is going to be game-changing. The other thing that I think is going to be game-changing is the access to the right knowledge at the right time. I don't think either on a personal or on a commercial basis, we truly have fathomed how having all of this knowledge in context, at breadth at our fingertips is going to impact our home and work lives because we're only just starting to scratch the surface of applications and to your point sam most of the people that we work with know this stuff but everyone else that sits outside of our ecosystem they don't live in these worlds and they don't have these conversations and they don't live and breathe this tech so i think it's going to be it's going to be the next couple of years i think it's going going to be as this layers into the real world and people use it personally and commercially, it's just going to be game changing and so much fun and exciting.

20:48I want to talk about very quickly before we move on about Alphabet. And they have just sold a 100 year bond, which is just incredibly, I mean, this is game changing stuff, isn't it mad? So they have just raised 20 billion, five more than expected. And looking at the pricing structure, it looks almost as safe as government bonds. I mean, that's what the market has basically priced it at. And it just feels like this is now pretty much kind of utility, too big to fail. This is infrastructure stuff. What else would you pick up on this 100-year bond from Alphabet? Lots of interesting things in it. But I think one angle in particular is what we've spoken about many times before, which is this slightly odd way we've structured pensions, in particular in the UK, where there's this notion of kind of matching assets and liabilities and kind of hedging your interest rates on a long duration basis.

21:47And what do I mean by that? Well, normally you'd say, you've got pension savers, and you're going to try and make sure that you should, must try and make sure that you have the money they need when they retire so they can get the pension they deserve. Now, if people retire in two or three years, you're going to want to go quite heavy into cash and bonds because you don't want money to sort of disappear and have massive stock market corrections right before people are trying to draw that money out. But then on the other hand, if people are retiring 20, 30, 40 years from now, or for those people that are retiring in 30 years, but then need to draw their pension over 30 years, that's kind of a 60-year window you're looking at.

22:30And if you have such a long window, equities are always best, right? Over a half century, equities have always massively outperformed bonds. And so you're going to want to invest more heavily in equities to maximize your long-term return, because that's really what's best for your pension savers. Well, in the UK, we have rules that say that the kind of the people that look after the pensions, the trustees, to be a good trustee does not mean to maximize long-term return. It means to minimize the short-term volatility and to minimize fees. And so what follows then is, you know, you basically are incentivized to lock up money for a long period of time in a bond, because that is what the laws and the rules are saying you should do.

23:23And it's hard to imagine anything that's longer time than 100 years, when it kind of exceeds most of our lifetime. So lo and behold, there is Google needing some cash, and here are this weird UK pension system looking for bond-like things that run for a long time, and it's a match made in heaven. But unpack it under the covers. If you have that kind of lock-in at 6 % over such a long time. Let's take the 1 billion pounds at 6 % over 50 years because this is a 6 % coupon bond. Now, at 6 % over 50 years, it gives you an$18 billion outcome, right? 1 billion invested, 18 billion out. It sounds good.

24:04But if you have 9 % instead of 6%, so equities instead of bonds, you don't get 18 billion. You get 74 billion, okay? So it's four times as much. And this is the idea of compound interest. It's very counterintuitive, right? We've talked about it as kind of the ninth wonder of the world, right? Einstein famously said it's kind of one of these things that are just magic. From 6 % to 9 % annual return difference, give you four times as much over 50 years. Now, our current rules are incentivizing the pension funds to lock in at the 6%. It just doesn't seem quite right. Now, on top of that, on top of that, where's the money going?

24:49It's going to fund a US hyperscaler. It's exactly the thing we've been talking about over the last year. It's not going to make the UK or the European economy grow better, faster, stronger. It's going to provide really cheap financing to a very large US tech company that's very dominant. I love Google, but I don't think they need to be subsidized by our money. But here's the kicker for me, Mads. Here's the kicker for me. is like you've got 35 trillion or so in savings across Europe. And all of these people that are putting their money into whatever investment vehicle, pension savings, whatever, they don't even know that it's going to the hyperscalers.

25:27They don't even know that it's going to go and feed the 100-year bonds and the beasts. No, and to some extent, I mean, all they should expect is for the trustees to invest the money to give them the best possible return. And that's just not what the rules are saying today. And that's what I find to be so frustrating, right? The rules should say maximize the long-term return. And when it comes to the tax incentives we're giving people, those tax incentives should be aligned with what is in the best interest of our economies. Neither of those things are true today. And I know Rachel, she keeps saying I've fixed it, I've done it.

26:02We have not. It's not. We have not really scratched the surface of this. Yeah, yeah. Also, it kind of raises the inflation challenge as well. I always wonder if you're, what's their coupon rate? Six and a half percent, I think, for this. It kind of makes me think, well, this is a very long-term bet on a lot of very, very interesting assumptions. We'll see where it lands. We'll see what happens. It's kind of put it into a whole new realm. I think the only other company I remember doing this, well, this is back in the 90s, was Motorola. So they had a 100-year coupon as well, didn't they? Do you know what happened with that?

26:36I didn't pay much attention. Do you know how that ended? a very nice frame on a wall somewhere i imagine a very like like shipping shipping bonds or i don't know what what are the what are the big old school things are there that are probably now framed pieces of paper in somebody's library somewhere with absolutely no value i want to talk about european sovereignty so some interesting things caught my eye this week that i'd like your thoughts on so pitch book kind of opened this a little bit not really it's kind of a line so stay with me their report stating the vc is a two-tier market so ai and other no great surprise it's slightly deeper than that as in there are now two vc industries i would argue there's the mega funds and then everyone else with ai and everything else but there were also some bits around every single eu lead like mers and von der leyen just sharing really kind of punchy we're gonna crush planning, we're going to, you know, if you're not on board with this 28th regime, we're going to set up a two-tier European Union.