ASML: Transition in 2024, Strong Growth in 2025 (FY24Q1 Earnings Call Minutes)
ASML released its first-quarter financial report for 2024 on the afternoon of April 17, 2024, before the U.S. stock market opened. The key points of the conference call are as follows:
For a detailed interpretation of ASML's 2024 first-quarter financial report, please refer to " ASML: Performance Collapse, Can AI Not Reach Lithography Machines? 》
I. $ ASML(ASML.US) Financial Report Highlights:
II. Detailed Content of ASML's Financial Report Conference Call
2.1. Key Points from Executive Statements:
- Market Conditions and Expectations:
① Memory Market: Demand growth mainly driven by advances in DRAM technology, especially in support of the conversion to advanced storage technologies such as DDR5 and HBM, is expected to lead to increased memory revenue.
② Logic Market: Customers are gradually digesting the significant increase in capacity over the past year, and logic revenue is expected to decline.
- Technological Progress:
① EUV Technology is expected to continue to drive revenue growth in 2024. The number of EUV 0.33 NA systems will remain the same as in 2023, with additional revenue expected from NA systems.
② The first NXE:3800E has been shipped this quarter, with productivity increased to 220 wafers per hour in its final configuration, 37% higher than the NXE:3800D, and is expected to become the preferred tool for advanced nodes in memory and logic.
③ The first INA or 0.55 NA EUV system has been shipped and is in installation; the second system has also started shipping. This system first emitted light at ASML's Veldhoven laboratory and achieved a new record resolution of less than 10 nanometers, with wafer exposure expected to begin soon.
- Outlook:
① Short-term: 2024 is expected to be a transitional year, with the company continuing to invest in increasing capacity and technology to prepare for market recovery; full-year revenue expectations remain unchanged, with the second half of the year expected to be stronger than the first half. 2025 is expected to be a strong year, benefiting from industry cyclicality, government incentive programs supporting new fab construction, and other factors② Long-term: Considering long-term growth drivers such as energy transition, electrification, and AI, it is expected that the demand for technology and material nodes will increase.
2.2 Q&A Analyst Q&A
Q: You mentioned that in order to achieve the target for 2025, we need to achieve an annual run rate of over 4 billion euros. Do we really need to rely on EUV orders to achieve these goals? And where do these orders mainly come from? How much of the memory orders are from China?
A: Order intake is indeed very unstable, which we have been observing for the past six months. The total orders in the past six months amount to 13 billion euros, 6.5 billion euros per quarter, which is quite substantial. When I mentioned the need to reach a 4 billion euro operating level to achieve the mid-point next year, this includes EUV. The order situation in the past few quarters and the most recent quarter indicates that some usual customers did not appear in the orders. Therefore, we need to pay attention to the plans of some major customers, especially the wafer fabs. These customers are planning to place significant orders in the coming quarters, which are necessary to achieve the mid-point.
Regarding memory orders and the Chinese market, we usually do not disclose the geographical distribution of orders. Although there are some orders related to China, the order distribution is more widespread, unlike the regional concentration of first-quarter sales.
Q: With the new regulations and rules in the Netherlands being introduced one after another, have you reassessed the impact on sales in China? Because three months ago, you mentioned that this could lead to a 10% to 15% decrease in sales. Do you have any updated views now?
A: The situation has not changed. The 10% to 15% we mentioned at that time was because we realized that we might not obtain the license to ship the latest generation immersion machines this year. Our view has not changed, and the rules have not changed. We still expect strong sales in China this year.
Q: Could you talk about the trends in your China business in the most recent quarter? Because the Chinese market has been performing strongly, there have been concerns that China may reach its capacity limit this year or later this year. What are your comments on the trends in China?
A: Relatively speaking, the Chinese market is still strong, but orders in the first quarter were slightly lower than in the fourth quarter of last year. Orders in China this quarter were 1.9 billion euros, compared to 2.2 billion euros in the previous quarter. Despite the decline, it is still strong, partly due to subdued demand in other regions globally. We expect the market to accelerate growth in the second half of the year, with a stronger performance in the second quarter. Although the first quarter did not break records, we expect the Chinese market to remain strong this year compared to other recovering regions.
Chinese demand remains strong, especially for mature technologies. We expect to increase capacity by 380,000 wafers per year. Although last year's capacity increase was slightly lower than expected, the Chinese market continues to perform strongly. They are increasing capacity to meet global demand, and market share will increase in the next few years, improving self-sufficiency. We believe that the reasonable increase in mature capacity in China aligns with expectations for demand over the next decade.
**Q: Can you disclose whether the large orders from wafer fabs and logic customers are for 3 nanometers or 2 nanometers processes? And between these two, how many levels of EUV technology are involved?**
A: Logic customers and wafer fab customers are still digesting some of the capacity they have built, mainly for 3-nanometer and 5-nanometer technologies. As for the capacity for 2-nanometer technology, it is expected to start large-scale promotion next year, which will be the next wave of EUV orders. We will now focus on the order reception for 2-nanometer, and the number of orders should come in the next few months.
In terms of layer count, we expect the 2-nanometer EUV layer count to be very similar to the 3-nanometer period, as 3-nanometer is mainly a equipment transition rather than a layer change. Therefore, our expectations for the 2-nanometer EUV layer count are the same as the expectations we have shared with you over the past few months.
Q: As we consider orders and plan to fill the situation in 2025, the company has clearly stated that it intends to pre-build low-cost tools this year and next year. Does this change your customers' view on the order pace?
A: I think that would be a bit opportunistic. I prefer to see our relationship with customers as a partnership rather than just a transaction. So, I don't think this necessarily will have an impact. Of course, as we have mentioned in previous calls, in order to create as much flexibility as possible for next year, we will do some pre-building, but of course, we will work closely with customers and do pre-building based on understanding their needs. Based on our interactions with customers, we have a good understanding of what they really need. So I think what determines our annual plan and our pre-building plan is more the interaction with customers, the confidence we gain from these conversations, rather than just whether orders will be received.
We will prepare inventory in advance because our delivery cycle is very long. We will convey this information to shareholders. Of course, customers are also aware of this, and then we will have final commercial negotiations for orders. Combining these two, orders may take longer, but this is normal.
Q: I would like to know the situation of memory orders you see. Is it still more dependent on HBM capacity expansion, or have you seen some budding demand for more traditional DRAM?
A: No, I think the vast majority of orders we saw in the memory field last quarter are still technology-related, mainly DDR5.
Q: You mentioned that inventory accumulation has little impact on delivery cycles. However, according to your statement last quarter, the delivery cycle of EUV is 12 months or longer. Now, based on the 2025 target, you need to reorder about 70 EUV machines to reach the mid-to-high-end level. After my calculation, we only have about 30 EUV machines in our order backlog for next year. This means you still need to order more than 40 EUV machines to meet the revenue target for 2025, and customers waiting until the last minute to place orders could be very risky. Can you explain?
A: We maintain continuous communication with customers to understand their needs. Orders go through some cumbersome procedures and negotiations, incurring costs, and eventually turn into purchase orders, some customers may therefore delay placing orders. But the fact is, we understand the needs of customers, and customers also understand our situationAs long as the needs of both parties are met, being open and honest, orders will appear. This is not a dangerous game, because we both need each other.
Q: How long is the delivery cycle for High-NA now? If we assume mass production will be achieved in 2026 or 2027, will there be more orders in the next few quarters? From a product and technology perspective, what key breakthroughs are you still expecting to drive wider adoption?
A: I think there are several important milestones. First, customers have already booked double-digit EUV High-NA devices without seeing any tool images, which shows their confidence and commitment to our technology introduction. Next, we successfully generated the first image with a resolution below 10 nanometers, which is a huge milestone for both customers and us, proving the effectiveness of the technology we have been developing for years. In the coming months, customers will visit our lab and start exposing their own wafers, which will help us better understand the next steps in the development of High-NA technology. These significant advancements make us excited about the future and will continue to drive our work forward.
Q: Considering the performance, power consumption, and cost requirements, how do you view the high-performance computing field becoming increasingly dominant? How will this affect your long-term view on leading process growth?
A: When considering high-performance computing, especially in the field of artificial intelligence, I have emphasized multiple times that artificial intelligence is data-driven, involving understanding the correlations between data and integrating with intelligent software. I also believe that industrial IoT will be the field for many artificial intelligence applications. To collect all this data, we need sensors. Mainstream semiconductors are crucial in this field as they involve sensing technology, but I don't like to use the word "mature" because it seems a bit outdated.
In fact, mainstream semiconductors are very critical, and their fusion with advanced semiconductors is essential for the growth of the entire industry. While cost is an important issue, as we create more value for transistors, the cost of the new generation of transistors will increase. The development of artificial intelligence and productivity improvements in some application areas will further drive the value growth of the next generation of transistors.
Therefore, we need to work together with customers to achieve the energy efficiency performance of the next generation of transistors, which will promote the demand growth for mainstream semiconductors. My confidence in the future comes from these observations and discussions, as all of these are interconnected.
Q: To reach the median in the next three quarters, approximately 4 billion euros in orders will be needed. If considering a higher range, an additional 2 billion euros or more may be required per quarter. The current order backlog is about 18 months, and I expect that by the end of 2024, some equipment in the order backlog will be shipped in 2026. So, is 4 billion euros the correct number for the next three quarters? Or should it be higher?
A: If considering the median, it is 4 billion euros. We have already excluded the part related to post-2025 in the calculation. As for your question, if considering the high end of the range, there is a difference of 500 million euros between 3.5 billion and 4.5 billion, divided by 3, approximately 1.7 billion euros per quarterQ: The US government has requested the Dutch government to prevent maintenance of certain aspects of infrastructure installed in China. What is your view on the potential impact? Considering that equipment cannot operate properly without service, can it be assumed that any ban may not cover all revenue from Chinese services?
A: Regarding the service issue in China, this has become a topic of discussion between the two governments, and we should let them resolve it. We are providing relevant information on the services we offer to them, but the ultimate resolution of the issue will still require government discussions, possibly involving national security interests. Currently, we have not encountered any issues preventing us from maintaining installed infrastructure in China.
Q: You mentioned that electrification could be a potential factor driving lithography demand. We have seen some news reports stating the need for technology to address strong EUV-driven grid demand and the demand brought by AI servers. Therefore, could you provide some background information on what this means for long-term lithography demand?
A: Grids, electrification, grid construction, investments in renewable energy, and investments in smart grids will be important drivers for mainstream semiconductors. Electrification involves mainstream and advanced semiconductors, not limited to a specific industry. Investments in renewable energy, such as solar and wind energy in the Netherlands, require upgrading smart grids to manage unstable power supply situations. This requires a large amount of mainstream semiconductors, and our current production is still lagging behind expectations. Therefore, global investments in mainstream semiconductors become more important as they play a crucial role in power generation, distribution, storage, and usage. This understanding highlights the importance of electrification in driving demand for mature chips.
Q: The first question is about the guidance for the second half of this year, clearly expecting accelerated revenue growth. Could you elaborate on the main reasons for revenue growth in the second half, especially in China? Do you expect the Chinese market to maintain a strong momentum in the second half of the year?
A: The situation in the second half is very consistent with the current orders, as we are already at full capacity for the entire year. The shipment plan covers multiple fabs opening and incremental expansions in 2024, both in logic and memory areas. For installed infrastructure business, we are conservative and expect growth in the second half to be slightly higher than the first half, especially in upgrade demand. As for the Chinese market, we anticipate strong demand to be maintained, although it may be affected by supply constraints in other regions.
Q: The last quarter saw a lot of strong orders in the memory field for technology procurement. Can you reveal when we can expect to see the subsequent impact of DUV orders for memory capacity? Interested in knowing what your customers think about this capacity situation?
A: Well, it is clear that as we mentioned before, memory utilization has been steadily increasing for some time. If this trend continues, as expected by the market and our customers, then in the second half of the year, you should see the memory market not only undergoing technological transformation but also increasing storage capacity. Therefore, this aligns very well with the developments we currently see in the market, and it is also a driving factor for the second half, with capacity expected to further increase in the memory marketQ: Besides EUV, what is the proportion of China in our orders?
A: We will not disclose the specific proportion. The geographical distribution of order composition is information we will not make public. We have mentioned before that China accounts for slightly over 20% in orders, so the sales proportion is roughly within this range, and there has been no significant change in this regard.
Q: Can you update us on the NXE:3800E? How is the throughput of early systems in the first half of the year, and what improvements can we expect in the second half of the year?
A: The full configuration will be launched in a few months. Therefore, in the second half of the year, we will increase its capacity to 220 wafers per hour. Initially, we will quickly ramp it up to 195 wafers per hour, and then in the second half of the year or early next year, the equipment about to leave the factory will achieve a capacity of 220 wafers per hour. As for the equipment still at 195 wafers, they will be upgraded to ensure that the entire equipment group can reach a capacity of 220 wafers per hour by early next year.
Q: How does this increase in capacity to 200 wafers per hour impact or drive orders exceeding 4 billion euros per quarter?
A: In fact, it does not. This relates to revenue recognition, not purchase orders, as the purchase orders signed by customers are for fully specified equipment.
Q: How much of the wafer capacity in the industry has adopted EUV technology? With the industry adopting EUV technology for DDR5 and HBM, do you think this proportion could grow to what extent? You have previously mentioned that over a longer period, possibly around 2030, the distribution of EUV technology between logic and DRAM could be 70:30. Do you think with the increase in layers in HBM, this proportion could become 50:50 in the next five to ten years?
A: It is difficult to answer the exact capacity question as we do not have all the information. But I can tell you about the trend in the DRAM field, which is that the demand for EUV technology is increasing at each node. Based on the roadmaps we currently receive from customers, we see the number of layers continuing to increase to meet the demand for DRAM.
As for high-bandwidth memory, although the equipment itself has not changed much, there is a significant change in the ratio between arrays and logic, usually requiring larger chips, hence more wafer capacity. So, while the layers may remain the same, the number of wafers required to produce high-bandwidth memory may increase. We are working to understand the impact of high-bandwidth memory on the overall capacity demand for DRAM, but there is no conclusion at the moment.
End of Document
Risk Disclosure and Statement of this Article: Dolphin Research Disclaimer and General Disclosure
