ASML avisa que la demanda d’IA continuarà limitada per l’oferta durant anys

We are meeting and talking at a moment where there is a lot of focus here in Europe on the decision by the US to impose export controls on anthropic models. And Europe is feeling the impact of that. And it's a reminder of the questions around sovereignty, particularly around AI. And you are pushing hard on this. What is your message to Brussels? What is it that Europe needs to do to ensure that it is not vulnerable to those kind of supply chain risks? Well, I think indeed there's a lot of discussion around that and sovereignty has become a key award because people realize that strategically if you don't have a big enough part of a certain ecosystem in this case AI you're going to be exposed to the goodwill of other parties But I think what's very important to understand when it comes to sovereignty is to get sovereignty you need innovation first you have to do things in the right order It's not enough to say I want sovereignty. What's very important is make sure that innovation can happen in Europe. And I think that's a bit the idea of the tech creators group we have created, where we believe that it is very very important to have a dialogue with the different government, Commission, but also national government, to help them to understand how to create the condition such that

the AI ecosystem can develop in Europe. You will never have the entire ecosystem, right? You cannot do it whole. It's too much. You cannot do it whole at the same time. But you need at least to have enough, overseas, skin in a game in a very unique ecosystem so that when those discussion or those this year in Japan, you have a choice. So where can Europe have skin in the game? They've got skin in the game with ASML and the role that you play. Where else can Europe have skin in the game when it comes to AI? Where is it? A very important place is the market itself. The European market is pretty attractive. You look at 22% of JDP. It's a very, I will say, a mature market, very good market. So this means we are user. We are buying a lot of stuff. And that's to start with a strange for Europe. That's the reason why Europe was created. Now, if you look at the other part between maybe a smell and that,

We have a lot of holes and sometime we discuss a lot semiconductor manufacturing. Yes, there is a gap there, but most probably not the right place to start because semiconductor manufacturing is only needed if you have people to buy semiconductor. Today if we had a 2 nanometer or a fab in Europe, most probably most of the wifers will go to the US. So since we have a strong market, We have to start by looking at that and focus on what we call the market demand product and have strong initial project around AI application around maybe AI product Then maybe I chip design and then chip manufacturing but you have to look at the entire sequence and look at it in the right order. How would you score Europe right now on a score of 1 to 10 in this AI race? Well, I think I've said that before, I think that if we compare how we do on the entire ecosystem versus the US and China, I think today the US is a clear winner. I mean, they are looking at champion across the entire AI semi-acquisist team. I think the one place that we're missing a bit out was manufacturing and I think there have been extremely aggressive in bringing some key company to manufacturing the US.

They can do that because they buy chips, 80% of the advanced chips, the manufacturer worldwide is bought by the United States. So I think the US is doing very, very well. China has been investing also across the entire ecosystem. I think most of the people did do very well on the application side, a bit less on what's before that, but I think they are driving the use of AI, most probably more aggressively than any other country. And again, when you look at Europe, a lot of horse. So I would say, you know, how can we quite be behind what's happening elsewhere today? So that gap is obviously there. How interventionists should European government be? The US taking a 10% stake in Intel. Is that a playbook that we should copy? Well, I think two things, maybe. So the first thing, we need to have a dialogue between the industry and

government. So government is to really develop instruments, regulation or lack of regulation in some cases to make the industry champion life a lot easier. I want to ask you about the IPOs that come through a SpaceX notably and possibly open a ion anthropic coming up into the future. Are we underestimating the capex spend that could come about as a result of that capital raising and what it means for the demand for your lithography missions? Well, if you look at the worldwide demand for AI infrastructure is still enormous and the demand for a AI edge compute start to come up. So the demand is continuing to go up. I think a lot of us are convinced that we're going to be looking at a supply limit in market for AI for semiconductor for quite a few years because the buildup of the infrastructure is huge. And I think some of the IPO you mentioned are just pointing to the opportunity that the eyes is represented. We are early in this still, where are we in that? I think we are early because even a year ago the AI company we are already very vocal about the need for this architecture, the need to build up capacity, the industry, the semi industry were still a bit hesitant to move.

and the move really happened basically at the end of last year and very quickly. And now we all need to build capacities so our customer have to build fab. We have to make sure we build enough equipment to fill those fab. And a catcher will take quite a bit. And that's only for the infrastructure. And then the application will come. Are you seeing a pull forward of all the other teams? We see a pull, we see customer continuing to create visibility, longer term visibility, telling us, well, this is a bit here to stay. So I would say that the trend we see is still very strong towards basically demand. Run with me on this one. Data centers in space. Is that is every potential additional revenue stream for ASML if, if, if, musk pulls that one off? Well, I think, you know, you need a certain amount of data center and they will be in space. They will be on the Earth.

doesn't necessarily change the number. I think the discussion about data centering space is to maybe address another potential bottleneck in the ecosystem which is energy. I think the discussion around putting data set in space is there because there is an idea that if you do that, the energy consumption or the energy of a very big deal will be there. So maybe not too much through with the total data centric capacity. but maybe more about trying to solve another potential big issue, which is an energy for a data center. Elon Musk is also looking at a terra fab, potentially one terra what of power. What's the potential upside for ASMR or not? Have you talked to Musk about that project? Is that a potential uplift as well for the business going forward? Well, you know, any added capacity of course, a potential upside for us.

We are looking very carefully on the development on when things will happen at what speed. New projects are opportunity as long as you're not supply limited. That's also important to say that. We have to make sure that this is really being taken care. The Tera Fab is an example of a major Fab project. If you look at some of the DRAM project in Korea, you also look at millions of whiffers. That's also pretty big. Okay, so Korea and the build out there is going to be a driver as well. Staying with the US though, one of the top house Republicans on the House China Committee sent a letter in April suggesting that SML was going to start shipping Duv the top of machines into China. There's some concern there. Do you have a response to that? No, it's not a response because we have a dialogue with every government. I think you know that in the last few years we have been subject many many times to those kind of negative remorse about what we do. Every time the remorse I've been quite, I would say potentially hurtful even for ice-a-mail. And I want to say it again, I mean we have been following everything around and we continue to do that. The world are changing by the way quite often and sometimes something you could do, you cannot do anymore.

But when this happens, we change our behavior accordingly. So we are extremely cautious in trying to follow or follow basically the rule of flows. It's very important. And that's true for DPPV, that's true for UV, that's true for any activity we have in China or anywhere else in the world. And I can promise you there's a lot of people in ACML making sure that we comply. Compliance is a very important I would say elements in a very difficult geopolitical environment. The Indian Prime Minister, the Render Ramonu, will be here in Paris tomorrow. Yes. You've partnered with Tata on their aspirations around boundaries and build out some of your conducts to capacity. How much of a growth story potentially is India for airs and Europe? So I think so first we're very happy. It was this partner ship.

because we are partnering at the very start at very beginning. So of course we all know that India today doesn't have manufacturing for the mechanical sector, but you know by 2030 they believe that 10% of any chips will be used by India and like anyone else that become therefore a strategic question to apply and therefore they want to get their factory. So first five will come up next year. A lot of learning has to happen and we will help India, Tata, for this partnership to accelerate the learning and you know as learning gets successful you go to your opportunity. So we're very happy to start now because we look at that as a growing opportunity for the next two years but for the next five and ten years in India and we're very happy that we are there from the start.