00:00It's kind of this contradiction, right? The headline is like really positive. Raising that guidance, raising that outlook. The demand
00:08is clearly there. The stock falls. And so like it seems like the expectations were high. People are already looking
00:15much further ahead in terms of that demand picture. What was your read on it?
00:21Sure. So that's a very good point. So basically, I would say don't forget that to start with. ASML is
00:27already up nearly 40 percent year to date. And semi-equipment segment has been one of the strongest, I would
00:33say, sub-segment in tech year to date in this year.
00:36So therefore, I would call the set of results reported by ASML today as solid. They raised full-year guidance
00:43by 4 percent. The Q1 gross margin was way above consensus expectation because of strong mix.
00:49At the same time, they confirmed that they are increasing their capacity to about at least 80 EUV tools from
00:57at least 60 EUV tools year on year in 2027. So I would say the set of results has been
01:02very strong.
01:03But at the same time, as the most, let's say, successful or high-profile tech company in Europe, ASML is
01:11always well-owned and ASML expectation has always been elevated.
01:15So I would call today's share price movement as just lack of revision on the back of this solid set
01:23of results instead of anything which is worrying.
01:25Tammy, can we just linger for a moment on what you just said? 80, 8-0 EUV machines. To a
01:32lot of people, they'll say, hold on, I don't understand that.
01:34This is the world's most important maker of machines in the chip manufacturing process.
01:39And they're guiding that they'll make 80 of them. Of course, the context is they're $400 million a piece. Just
01:45explain why that number is significant.
01:48Sure, of course. So those tools, you are correct, are sold for $230 million each.
01:54So those are probably one of the most expensive set of tool in the market or in the world.
01:59And I would say that it's important because we are moving from 60 in this year and we were at
02:07about 40 a few years ago.
02:08So therefore, from 40 to 80, they're already doubling that capacity they can serve the market, which is showing that
02:16AI demand and memory demand has been super solid over the next, let's say, six months or two years.
02:23And going forward, it's likely to support the further expansion of AI market or memory market.
02:30So therefore, as a result, I would say that 80 number, it doesn't look like a big number.
02:35But don't forget, those tools are $230 million each.
02:39And we were at 40 two years ago.
02:42Are they in any way supply constrained here, Tammy?
02:45I mean, some on the street hope to see the number 90 thrown around.
02:49Is there a limitation factor to them?
02:51Could they be making more?
02:54So in our view, we don't think that ASML will be a bottleneck.
02:57You are right that 90 was mentioned previously as the potential capacity number.
03:03But ASML has been super careful in terms of building capacity as per customer demand, because you don't want to
03:09end up in the situation that you built 90 tools or whatever the number is, but there is no demand
03:14for that.
03:15So therefore, as a result, I would say that 80 number is a result of ASML's discussion with customers.
03:21And that 80 number, for the time being, is a reflection of ASML's agreement with customer how to serve the
03:28undermarket demand.
03:30And I don't think it's a limitation in theory.
03:33In my view, ASML can actually do more if the customer need more.
03:36And what ASML has been needing is just to hire more people, expand with their supply chain.
03:42So I don't think ASML will be the bottleneck.
03:45Let's talk about the demand constraints or fulfillment, because it feels as though it's insatiable in terms of need for
03:53chips and therefore needs for equipment.
03:54But they had exposure to China.
03:56How has that continued to unfold?
03:59So from a China perspective, China used to be 40% of their business two years ago.
04:04And now with the non-China business being growing strongly, China is about 20% of the total business.
04:11And China has been stable, I would say.
04:13China was slightly better than they previously expected.
04:16And China will be stable in this year.
04:18But in this quarter, the four-year race was purely because of non-China business was stronger than expected.
04:24So going forward, I would say ASML will be continued to be driven by this non-China business AI investment
04:30memory shortage to further continue as the case for the next few years.
04:37And China, I do have the view that China will be flattish.
04:40And China is working on their localization, of course.
04:43So the demand for China is not likely to fall off the cliff.
04:46So therefore, as a result, in conclusion, over the next few years, ASML will be driven by both factors, which
04:52non-China business will likely to be the key driver.
04:56Talking about $230 million for EUV.
05:00I mean, if you think about TwinScan EXE, some of the people I talk to, you go sub 2 nanometer.
05:05They can get $350 million for that.
05:07We'll get back to it at another date.
05:10There's a big project.
05:11It's called TerraFab.
05:12It's basically on paper, but it says that in the future, there'll be one terawatt of compute capacity put through
05:21FABs operated by Tesla, SpaceX and XAI.
05:24That would on paper make them one of the biggest buyers of ASML's equipment.
05:29Have you and the team modeled for that being real?
05:32Or for you, is it still an idea?
05:36I mean, we would love to.
05:39And you're absolutely right.
05:41So ASML do have a version of a machine which costs $350 million, which is called HiNA.
05:46That probably will be seen when TerraFab is ready to take equipment.
05:51So for the time being, we haven't put TerraFab in our numbers.
05:56But as per our understanding, the intention is real.
06:01And there has been discussion already within the supply chain about TerraFab.
06:04But for the time being, it won't be something happening in the significant way from a number perspective over the
06:10next two years.
06:11So therefore, we haven't reflected that.
06:12But I do hope that we can put that in numbers in two years time.
Comments