Europe’s AI story. With Philip Lane

0:00There is a common narrative about AI. The US is in the lead, building data centers and chewing through chips at a crazy pace. China is moving fast too, coming up with seriously impressive AI models. Other Asian countries are heavily affected as South Korea and Taiwan pump out the components, powering the boom. And Europe? Europe is being left behind, slow, plodding, saying it wants to catch up, but so bogged down in red tape that it falls further and further back. But is that right? And even if the relative position is correct, is AI making a difference to the European economy, to growth, inflation and investment?

0:46This is The Economics Show with Samaya Keynes. I'm joined today by the person you want to be talking to about Europe's economy, Philip Lane, Chief Economist of the European Central Bank. Philip, hello. It's nice to be back. Okay, so you've been on the show before, so you know what's coming. On a precise and objective scale of 1 to 10, how important is the AI boom for the euro area economy? I'm going to give that an 8, because I know you like sharp answers. And the 8, I think, is a mix between the fact that I do think it's having a visible impact right now. But again, if you look five or 10 years out, then I think that eight will be validated.

1:27I do love the high number. Much appreciated. Okay, so, but let's put some detail on that. Where are the main areas where it is showing up? So I think, let me divide AI into four dimensions. Who's developing the frontier models? We have Mistral in Europe, but of course, most of them are in the US or in China. Number two is compute. So where are the data centers and so on? US has most of the compute, a very big lead there. Europe is growing, but from a low level. But the two other areas are the global supply chain for AI. You mentioned, of course, Korea, Taiwan have a very big role. But European firms have a very big role as well.

2:10So if you think about ASML, Zeiss, Schneider, this is an era where it's quite visible. So the exports of AI-related equipment has been very strong in Europe this year, and we expect it to continue. And then the other area, which in the end is the most important, is AI adoption. And we're seeing, I think, good AI adoption rates by firms and by workers. There is, of course, a debate about who will capture the benefits from AI. Is it going to be the adopters or the creators of the models and so on? One of the pieces of research, by the way, at ECB has been, regardless of what any one firm thinks, if they think their rivals are adopting AI, the kind of FOMO dynamic is there.

2:56Even if your own inclination is, well, I'm not so sure, you can't afford not to be adopting AI. So I think the adoption rates, they vary by sector, by firm. But the surveys we run show, I think, a fairly good pace of AI adoption. Can I just go deeper on trade for a minute here? Obviously, you do have European firms that are contributing to the AI build out. Are you seeing this in exports and imports? Which categories? What about prices? Right. So it's good that you kind of list that out because often we think about trade volumes. But of course, this year, what's been very visible is a sharp increase in the prices.

3:38So if you're producing something that's in very high demand, the firms are planning capacity increases over the next several years. But right now, that will find an expression in terms of high export prices. So if you like, that's good for Europe when European firms are able to sell their output for high prices. Europe is also importing, but let me emphasize, what surprised us this year is the outperformance of exports compared to what we were expecting. Okay, so exports have surprised. On imports, is there any difference in what Europe is experiencing and what, say, the US is experiencing? No, I think that's a good question because I think we should be clear at this point, most of the kind of leading-edge massive data centers are in the US.

4:26So data center construction is picking up in Europe, but we're not seeing those kind of really intense stuffed with NVIDIA chips type data centers being built at any or close to the same rate. So we don't have the import surge in that sense that the US is. How do you think the AI boom is affecting Europe's imbalances with the rest of the world? So this is, I think, an important question, which I've tried to highlight in the work I've done on this. because we know right now the investment is picking up in Europe, but investment is picking up even more strongly, for example, in the US. We also know that Europeans are funding the US investment boom in the sense of, you know, Europeans are buying US equities, the bonds issued by the US hyperscalers.

5:17So you might say, well, this is a very natural way the world should work, that the resources should flow towards the country doing the kind of leading investment. So you might say in that world, you would see a bigger US deficit than you would expect and a bigger European surplus than you would expect without AI. Under the hand, going back to the fact that adoption rates in Europe are, I think, decent. We're importing a lot of AI, if you like, IP from the US. So that's the software licenses when Europeans are using this technology. Yeah, exactly. So software license, I mean, the industrial organization, how it happens, some of it is raw in terms of software licenses.

5:59It could be American multinationals who are more choosing to, if you like, within the firm, have a transaction between the US and their European affiliates. And then their European affiliates are kind of selling the licenses to the European customers. But taken together, it does mean a rise in service imports from the US. And we have seen a visible drop in the European current account surplus in the last year or two. And this is playing a role in that. Okay, so if you're President Donald Trump and you're angry at the European surplus and the US deficit, actually, maybe this AI boom is making you happier.

6:39I'd say because you have these forces pulling in other directions. I mean, it's a sign of the success of the US AI industry that they're pulling in so much resources. It's also a sign of success that their revenue in part is but through exporting intellectual property to Europe and around the world. Okay, one thing we haven't really spoken about so far though is inflation. The AI boom isn't so much of a big deal there, right, for the euro area? Well, the headline is, of course, energy inflation is the biggest issue. But if you kind of drill down a little bit, it is visible to some extent. So within the category of goods inflation, then if you say, okay, what consumer categories also use semiconductor chips and so on?

7:28You can think about mobile phones, you can think about laptops, you can think about cars. Cars use a lot of chips. So we are seeing a bit of an increase in goods inflation. Partly that's energy, partly it is this AI spillover. Is it possible to put a kind of more concrete number on that? So when we look at goods inflation, it's gone up from about 0.5 to low ones. And about half of that, about 0.3, is coming from electronics, which I think it's a fair reading is a lot of that is, if you like, competition for chips and other components from the global AI boom is pushing up prices. I mean, you read that when you see narrative from the phone makers, the laptop makers, why they're raising prices.

8:14I mean, is there any world in which that could sort of spill over? In the US, you hear about the energy story, maybe data centers increasing competition for energy, is that pushing up energy prices for households? Well, this is the flip side of if the investment boom in Europe is more more moderate than the investment boom in the US, you don't have the same calls on energy. I mean, it's like in Ireland, we have a lot of data centers. It's visible in electricity demand in Ireland. So by and large, I think European countries are essentially trying to balance it in the sense of the more firms want to build data centers, the more I think the direction of travel is that they need to add their energy supply more or less in proportion to their increase in energy demand.

9:01Okay, so that's inflation. The labour market. So I think globally, the narrative is it's actually quite hard to pick up big effects of AI on the labour market. Maybe you're seeing a bit in terms of entry level jobs for younger people, perhaps in certain sectors more than others. Is that roughly the story in Europe or is it something more complicated? Right. So I think around advanced economies, we have this kind of low hire labour market. So unemployment is low in Europe. We're not seeing AI lead to, if you like, shedding of labor. Now, I think this isn't a shadow of when firms look ahead over the next decade, there's a big retirement wave coming in Europe.

9:40So if a company is looking ahead and saying, OK, if I do have to make an adjustment vis-a-vis AI, how am I going to adjust it? One is less hiring. We are seeing less hiring. and then also in terms of planning for future retirements. Where I do think, at least from my point of view, I infer it from the data, we are seeing fairly moderate wage increases. And I think in Europe, there are two forces helping workers in a lot of industries understand there's an upper limit to what's sustainable in terms of wages. So in a lot of sectors in Europe, it is competition from China, but also AI. So I think we've seen over a long period of time, unions in a lot of the EU area prioritise job security over wage increases.

10:30And the AI risk is, I think, contributing to moderation of wages. Okay, so the AI risk is maybe making people feel nervous about quitting or nervous about demanding a huge pay rise. Right, because remember, what's interesting, maybe different to the UK, is a lot of the wage negotiations in Europe are like for three years. So even if right now it may not be an immediate issue, if you're making a three-year wage negotiation, then each year it becomes more plausible that firms have an automation option. Right. And I mean, that potentially speaks to the much bigger question of who's going to capture the gains from AI, right?

11:10You know, I think economy-wide, there's this big question of, you know, what's going to happen to the share of income going to labour, going to workers versus capital, business owners. And if you're saying that we're maybe seeing this dynamic already where workers are anticipating that their jobs could be under threat and that's kind of encouraging them not to be so strident when asking for wage gains, maybe that suggests that actually it could push that labour share down? Well, here I think I have to go back to the trade question we talked about. So if you think about the European production function, it's European capital, European labour, but also imported materials of flack.

11:48And one of those imports is AI services from the American firms, also the Chinese firms and so on. And going back to the debate about where will the frontier firms make their money, implicit in that is a lot of the value added of Europe would have to be paid out to the frontier firms via basically a higher import bill. So the kind of closed economy version of labour versus capital, when you have an international economy, you also have to think about the other element. I mean, as you know, in these years, the other big import issue is, you know, how much we're paying out for imported energy. I think it's very important.

12:33In a lot of European discussions, yes, there's always an issue with European labour versus European capital, but the role of imports in production is a very big issue both on energy and now with AI. Okay, so yeah, in the US I think they don't tend to think about those international factors in that debate. Well, they're on the other side of it in the sense of, you know, as an AI exporter and also as in these days an energy exporter, it's the mirror image. So all those American lefties worrying about the labour share could be worse. You could be worried about value also going to foreigners. essentially yes this is a big difference here now I want to talk about investment right so you know this is an area where AI is showing up I think quite significantly in the euro area how are we measuring the AI effect on investment so the ECB staff have you know I've done a good job in in the speeches I've done this year I've been showing this kind of a digital investment proxy where they basically add up reasonable categories from the more detailed data to try and get to something.

13:43And it's gone up by about 70 % in real terms since 2014. And if you look over the charts I've showed, it is accelerating, I think. So this is significant, but let me emphasize, so far the contribution of data centers is pretty small. So a lot of this is investing in software licenses, essentially, and also equipment. I mean, even without AI, there's a more general digitalization trend in the European economy. So what I would say is one dimension is comparing Europe versus the US and China. The other dimension is on its own terms. And on its own terms, the counterfactual where we didn't have a driving investment, I do think we'd have a much quieter investment environment in Europe at the moment.

14:32And from the surveys we do, the firms are telling us that, you know, without AI, they would have scaled back investment plans more. Without AI, you wouldn't see credit dynamics being at the level where they are. And I think from our point of view, this is the most important issue is for the European economy. Is it having the European economy grow more quickly and also building the foundations for the future through investment? And that's where I think it's important to say it is. Okay, but you just mentioned credit, right? So how is AI affecting credit dynamics exactly for European firms? So we are seeing the banking system respond.

15:13Credit is growing around 4 % at the moment. It probably would have been more like 3 % without AI. So about a quarter of the growth this year in credit is coming from AI. even though those AI sectors are not a quarter to the economy. So credit, if you'd like, you might say, okay, credit is doing its job. But the counterfactual is, what would we like to see? I would prefer to have seen a world where the other type of finance, risk-absorbing finance like equity, would make a stronger contribution. If you want to be responsive to new possibilities, new technologies like AI, the responsiveness of the banking system has been there.

15:51but it doesn't cover, let's say, an AI native firm. Some firm that, if it had enough equity finance, could grow quite quickly, could basically be a mechanism to roll out AI across the European economy. Do you get the sense that when it comes to this AI build-out, Europe is investing enough? I mean, there's a lot of data centre activity, if you like, in the works. So going from permitting, planning, to actually putting the shovels in the ground and so on, It's going to take time. So I would say the private sector, the commercial response is there to some extent. But in terms of should the policy environment do more to say, well, actually, let's be more ambitious here.

16:36To some extent, it's there in some of the European policies. But I would say, going back to my score of eight at the beginning of this conversation, Europe needs some clear strategies to boost the growth rate. and I think working out what is the infrastructure that can boost the European growth rate. Building the infrastructure for AI is important there. Well, we've just had the US midterm results, right? And there's a lot of discussion about whether maybe data center construction in the US will be halted. So maybe there's an opening. Maybe Europe can be bidding for some of that capacity that isn't being built out so fast in the US.

17:13Well, I mean, there's also debates here in Europe. So I think, I mean, maybe going back to the full AI issue, what's been interesting over the last year or so is more and more people pointing out that the speed at which this should go should take into account societal constraints, political constraints, labour, that the fact we have to recognise labour markets can't move that quickly. And so if we end up with maybe a slower pace of data center construction, if that creates the time to build more societal consensus in the US and in Europe and elsewhere, that may not be a bad thing. Okay, well, I am going to cut to a break now.

17:53And when we get back, I want to ask how AI has complicated the life of a central banker.

18:33We'll see you next time. CapitalOne.com slash bank. Capital One N.A. member FDIC. Hey, we're Nick and Jack, hosts of The Best One Yet. Behind every episode, there's research, email, spreadsheets, and our shared Google Doc, the headline hammer. With Google Workspace with Gemini, Google's AI is built right into the premium versions of tools we already use, like Gmail, Docs, and Sheets. No new tools or extra logins, just personalized AI for the way we work. That means less time on the work, around the work, and more time finding the stories worth talking about. Get AI tools for work. Upgrade to Google Workspace with Gemini at goo.gle slash workspace with Gemini.

19:10That's goo.gle slash workspace with Gemini.

19:18Okay, we are back from the break. We've been talking about the effects of AI on the economy. And your job is to manage inflation in the euro area. And so far, the AI effect on inflation hasn't been massive. But you clearly think it's worthwhile studying this revolution. You've dedicated several speeches to it. So how do you think about AI when it comes to monetary policy? Why does this matter to you? I mean, honestly, if we end up in a world where productivity growth really gets a lot stronger, then there will be a debate about the inflation effects of that. So for monetary policy, where it's showing up now, is the upward movement in long-term interest rates.

20:01So whether you look at the 10-year rate or even further out, when we look at that and you look beyond the discussion about the current cycle to do with energy price and so on, if you look five years out or 10 years out, what we see is the real rate, so the inflation-indested rate at that horizon. And to some extent, if you look about what do people expect, it's inching up a little bit in terms of the range of possibilities. So the price, not just what they expect, also what might happen. And in terms of what might happen, the markets are probably seeing a sharper increase in real rates over, let's say, the early 2030s.

20:43that is consistent with AI productivity more visibly picking up maybe around that horizon. Maybe by the early 2030s, we'll be in a kind of a phase where the gains, the payoff, if you like, in terms of productivity growth goes up. And, you know, a lot of people would say, well, maybe in that world, real interest rates will be more naturally higher. So that's a conjecture for now. but when you look at long-term interest rates, conjectures are priced. And so for now, that is leading to a tightening of financing conditions. So it's harder for companies to get credit. Right. But the ones who are confident that they will be obtaining some of the benefits of that productivity, in particular the hyperscalers, they are borrowing at that higher rate.

21:36So given that the hyperscaler activity is more in the US than it is here, from our point of view, we have the higher rates. We don't have the current scale of the investment boom the US has. And so on net, that can be viewed as a tightening of conditions in Europe, which will slow down investment across all sectors, including, by the way, firms which might have a good AI adoption idea or native AI firms. That is a consequence. Okay, so you're describing this quite cruel loop, right, which is that, you know, anticipation of future gains from AI, future productivity increases, potentially raises long-term rates, that tightens credit conditions for firms today.

22:22And if you are trying to borrow to implement your big AI investment, actually, that could make it harder. Right. So it leads to this differentiation. So the firms like the hyperscalers, which actually have a very strong balance sheet anyway, they can continue, but other firms have to adjust. So maybe going back, I mean, globally, higher rates now in the context of the global AI boom, also in the context of many governments also continuing to want to borrow a lot of money is leading to this upward increase in rates. but that by and large is not matched in Europe by the current level of the expansion.

23:02So I'm optimistic in Europe that we are seeing good AI adoption, but in terms of the capacity to roll it out, it is being held back to some extent by this upward shift in long-term rates. Okay, one thing I have noticed is that AI is increasingly being used to analyse central bankers. Have you also noticed this? And how do you feel about it? It is a bit destabilizing. So the ESP staff are hard at work running speeches, statements and so on through AI. So I think we have to do it. We have to understand how AI is interpreting what we say. So specifically what we're talking about here is, you know, you give a speech and then some AI agent...

23:49Summarizes it and so on. Summarizes it because you're a hawkish, dovish school. I know, I know. So we have to recognise that that's happening. On the other hand, I still really want to write my speech carefully. I want people to read my speech. I don't want people to read the AI summary of my speech. And I would encourage everyone listening to this podcast that there will be value in actually reading. Every sentence is carefully crafted. And just looking for the summary would be a mistake. Okay, well, I encourage podcast listeners to listen to the whole podcast rather than a summary of the podcast.

24:24All right, one last question. This is an interesting time for the ECB, right? Not least because there is an imminent change in leadership. So you know both candidates to be the replacement for Christine Lagarde, Klaas Knott and Pablo Hernandez de Coz. And don't worry, I'm not going to ask you to make an enemy of them by asking you to pick their favourite or, you know, give them a score on 1-10. But, you know, given your experience of working at the ECB, could you just share what you think the important qualities are for the next leader? What matters? So I think what's important for the ECB is continuity.

24:59And both Pablo and Klaas have been recent governing council members. Very importantly, I think they're a part of the strategy reviews we've done. I mean, I think the ECB operates on a strategic basis. I think we do have, I think, settled strategic priorities in terms of how we do monetary policy. And I think that's important that the incoming president, you know, builds on that. Of course, there's always room for a new president to put their own stamp on the organization. And of course, President Lagarde, I think, has been a very good leader for the ECB. I've very much enjoyed working with her.

25:38I admire her a lot. But it's also, I think, the case that any new leader will bring their own qualities to the organization. Okay, so change and also stability. Exactly. Perfectly clear. Thank you so much for joining me, Philip. It's been my pleasure.

26:01That is all for this week. You have been listening to The Economics Show with some air canes. If you enjoy the show, then I would be eternally grateful if you could rate and review us wherever you listen. This episode was produced by Katya Konkova, Original music and sound design from Breen Turner. Our executive producer is Manuela Saragossa. I'm Samaya Keynes. Thanks for listening.

26:44Vantage Score uses 400 % more credit report and alternative data than legacy models, bringing more people into the credit economy, like 5 million more mortgage eligible people. Banks win and francs win. Lenders win and so do Linda's. When you build credit scores the right way, everybody wins. Get started at VantageScore.com. Hey, we're Nick and Jack hosts of the best one yet behind every episode. There's research, email, spreadsheets, and our shared Google doc, the headline hammer with Google workspace with Gemini. Google's AI is built right into the premium versions of tools. We already use like Gmail docs and sheets, no new tools or extra logins, just personalized AI for the way we work.

27:23That means less time on the work around the work and more time finding the stories worth talking about. Get AI tools for work. Upgrade to Google workspace with Gemini at go. That's goo.gle slash workspace with Gemini. That's goo.gle slash workspace with Gemini.

28:02Stay briefed. Find Barclays Brief wherever you get your podcasts.