Transcription
So, on the question of A.I. and its impacts on the banking sector, how do you see A.I. reshaping the banking landscape?
Well, I think it comes back to the vision that we had on that ten years ago, which was very basic. It said in the future, it will be a digital financial system. It will analyze your spending, let's say your mortgage. And it will say, like, that's too much. I found a better offer for you. The only thing you need to do is say yes, and I'll do all that, all the paperwork and get it all done for you and save you the cost. Right. So to us, this was basically what we believed retail banking was heading. And a little bit like self-driving cars, don't know when, but no way that it will happen. And and that's the future. And it has profound impact and the consequences for the banking industry because, to a large degree, banks today don't need to be competitive because people don't want to switch. And it's difficult to switch and they need to bring all that data and all that stuff. So they don't need to be that competitive. They don't need to be that great. And this would mean that in the future, customer mobility would be ultra high. People would switch to whoever gives them the most value for the least money, and it's going to drive down the excess profits in the retail banking industry.
Where do you think legacy banks are most vulnerable then, Sebastian, when it comes to A.I. disruption?
Well, I think the most legacy banks will probably become pure balance sheets where, basically, you know, you have a big balance sheet and you optimize for return on equity, and that's pretty much it. I think the challenge is, and this was already part when we were, you know, we bought a company back in '11 that was building a front-end, kind of a digital-first generation of a digital assistant. And as a consequence of it, you could do payments without logging into your bank. And banks hated it because they realized it's so important. Time. If that assistant becomes so good, you're never going to use your bank. And then you're going to be fairly, you know, you're not really going to care if it's Barclays or if it's this or if it's Big Bank or whatever the assistant is. What matters. The person who provides you the value is what matters. The rest is just, you know, zero excess profit. And so I think that the what you have to decide is either you want to be a balance sheet, and that's fine. You can make money being in a balance sheet, but you would expect profits to be close to perfect market. Or you can be the digital financial system and you can try to provide that value, which is a difficult thing to be. And fewer probably can succeed, but obviously could be where the future of value of banking really is.
What would you say to those, though, who would say, look, the legacy banks, they have the datasets, they have the resources, then in fact, rather than leveling the playing field with fintech, A.I. is actually going to give a greater advantage to legacy banks because of those factors?
Well, look, I've always said that in my opinion, there are three players of companies that can play in the space, right? It's the big tech companies, Google, Amazon, etc. It's fintech, or it's banks. There will obviously be incumbents that try to, you know, reinvent themselves, be forward-leaning, apply to technology and so forth. And some of them may be successful, but they will also be fintechs. And in general, if I look at it today, I would argue that companies like ourselves and our Revolut and others, because we already are banks and we've kind of got an across that, you know, being a small startup, we're not small sponsors anymore. We're fully regulated entities with, but we have modern tech stacks. We don't have, you know, the COBOL code running in the back mainframe and all of that stuff. We don't have, you know, all of those challenges that big legacy incumbents have. So I would argue that our odds are pretty good at winning a big market share. But of course, some incumbents will manage to reinvent themselves. But I think Jamie is on to something really exciting. He's trying to do it now. You know, Solomon was trying with Goldman Sachs, but then kind of Marcus, you know, was thrown out when the fintech valuations dropped down. A lot of banks stopped their aspirations in fintech, and now they're waking up to that. FinTech never died. It was just a it was just a valuation issue. And 2021, 2022, the valuations dropped, but fintech was growing, the revenue was growing, the customer adoption was growing, it was just investor sentiment that changed for a few years. And now it's going to come back because now you see new back, you see revalued, you see all this stuff happening, you see Klarna, Affirm, etc. and you're going to be like, Wow, you know what? Actually, the disruption is happening.
And as you know, because you have that license, of course, in Sweden, that banking license, banking and trust, they go hand in hand. And there are still questions about trust when it comes to A.I. in terms of sometimes bias around decision-making on lending, for example, decision-making biases that can be baked into some A.I. components. How do you think about addressing trust and A.I. when it comes to banking?
I think in the end, like you obviously. Right. And we actually don't use much of LLM I, I on for example, on the writing. We still don't use machine learning. We don't use A.I. in that particular regards due to some of the risks and issues that you highlighted. But there are other areas where A.I. lends are extremely efficient and helpful. But I, but you're totally right. But in my opinion, it's very clear to us that as this technology progresses, you'll find ways to avoid the hallucinations or the issues of trust. And in the end, what we've seen, for example, with our customer service is that initially, we take dispute management as an example, but we've had huge success with the consistency in which an NLM takes the evidence of how to resolve disputes in information from merchant, information from consumers, and then decide what is the right thing to do, give the money back or, you know, let the merchant keep the money and so forth, things like that. The consistency of such decision-making is actually of higher quality and more consistent when performed by humans to review these decisions and they look at them. And if you ask the same, if you asked the same job to be performed by a human, over time they get a bit bored. It's not that interesting, or they may not be as consistent in the performance. So there's you can build a lot of trust by having an A.I. perform specific types of tasks because of the consistency and quality. But it requires some engineering and some thoughtfulness in how you set it up to make it work with some of these changes.
Then, Sebastian, I know you think deeply about this as well. How do you see financial services, the workforce, when it comes to financial services changing in the next 5 to 10 years?
Well, I for one have been trying to be a little bit more honest about this because I feel a lot of my tech bros are being slightly, you know, not to the point on this topic. I think there is a massive shift coming to knowledge work. And it's not just in banking, it's in society at large. There are 8,000 people in Brussels that work as translators, and I would argue that most of that can be done by A.I. today already, even though obviously some quality assurance and so forth would still be valuable to have humans to review. But the point is that you don't need as many as we did. So there is a massive shift coming. I think society will have to figure out what are we going to do because yes, new jobs will be created, but in the shorter term, that doesn't help the Brussels translator. He's not going to become a YouTube influencer tomorrow. So you may find other ways in how we support people from a society perspective. For us as a company, we see we have increased our revenue per employee from 400K to over $1,000,000 per employee in the last two years. We haven't hired. I mean, we've a little bit higher small recruitment, but generally we've shrunk in the company from 7,400 people to 3,000 now. And that's at the same time, while the revenue and number of customers has grown significantly. So we are simply not recruiting, which means that we can avoid being doing layoffs. And we have taken a lot of the savings in payroll costs and reinvested out into an acceleration of the compensation of our employees, which means that they are seeing a huge benefit from our applying A.I. internally. So I think that there is a path and how you can make this work both from a society perspective and a company perspective. But in the end, our job is here to drive as much value-added as possible price to our customers. And so like we need to use these technologies to accomplish that. And then in general, I think there will be implications for a lot of knowledge-based work.
Hmm. Do you worry? Clearly regulators are thinking about the impact on jobs as well. Maybe they should be thinking about it a bit more. Do you worry about regulations? Do you worry about the regulators stifling some of this innovation, Sebastian? Or do the rules being laid down provide clearer guardrails?
In Europe, there's a huge risk, right? Because at least for a while there was like going into total opposite and wrong direction. Now it's shifted a little bit, which is good. But that is a risk. I think that like, you know, we have to in my opinion, it's not like I don't see risks or challenges with A.I. and concerns for society. But we also have to remember there are non-democratic countries that are pushing the envelope on these technologies. And I rather see that democratic countries, countries are ahead of non-democratic countries in regards to this technology. So I just want to be mindful that we don't stifle innovation from that perspective. But with that said, currently, I think it's it looks more favorable. And I would still argue that there are other is going to be other consequences if you, for example, create true customer mobility, which hasn't existed in banking for a long time because we're all stuck with the banks that we have. But if you create that, you're going to, for example, have more customer mobility means funds and deposits can flow more freely between financial institutions. It's going to be a different world for banks to figure out how do you, what does that mean for, for example, you know, potentially if people move all your balances away, that that potentially has implications for your liquidity risks and stuff like that. So there's going to be new challenges, but in overall, it's good for society. It means less excess profits in what has been known not as well-functioning markets as they could have been.
Do you think, do you think A.I. leads to consolidation in the fintech and banking space, or does it open it up to more entrants?
No, I do think it leads to consolidation because basically what happens is that the cost of producing software is going towards zero, right? We used to have tons of people, very smart, very well-paid people sitting in rooms and typing out character by character, every line of code. And an institution like Klarna will have millions and millions of lines of code. Nowadays, that code can be generated by a machine, and that machine is getting better every six months of generating that code. So the cost of manufacturing code is coming down very, very quickly, which has to lead to the fact that excess profits in software technology will be less than it is today. And so I think it has material consequences. And what do you do as a company? Well, what Klarna has been thinking we should do is let us become larger. Let's become a global financial technology company. You see Revolut also aspiring for being a global retail bank. It's the same thing because you realize that you have to be at scale. You have to be super large client, actually by number of customers. Today, with 111 million customers, I would argue we already are one of the largest banks. Now, obviously, those customers doesn't necessarily bank fully with us just yet, but that has to be the aspiration because you just have to have more scale when you know there will be less money per customer. That's pretty much 110. But we tell our customers, yeah, yeah, except. But then what we tell our investors is, look, there's going to be a smaller pie, but hopefully Klarna will have a bigger piece of a smaller pie and it will still be a great, great return for our investors.
Yeah. One so far, investors are taking taking your word on that. Buy, buy, buy the moves in the share price. Sebastian, before we let you go, I need to ask you a personal question. How, how much and to what extent are you using personally in your own life outside, maybe outside of work?
Well, I use it all the time. And actually, since May, what I did pick up was vibe coding. So, I mean, vibe coding and exploring our code base because I'm not an engineer by training. And I must say, it's been mind-blowing to me. It's been fantastic. And I spend now, you know, my wife is complaining because when when the kids go to sleep, I'm like, Hey, can I just go and vibe? So she's not too happy with me currently. But it's just like it's so intriguing. It's so fantastic to learn these technologies. I think people should not be aware of technology. Which if you're worried about a technology, embrace it, learn it, educate yourself, find out how to utilize it, and you know, and then you'll be more, less worried because you know what you're actually facing, right? Otherwise, it's it's the things that you don't know that you that your people get nervous about.