2025 Annual Report

Prioritizing problems and executing solutions

Edison’s day starts early at 7:00. His first task is to review the pass-down report – a detailed summary of customer problems and the actions taken in the last 12 hours. Because customer support is a 24/7 operation, the night team ‘pass down’ their responsibilities to ensure continuity of service. Edison collaborates closely with first-line engineers, identifying the highest-priority issues and allocating tasks to ensure each challenge is handled quickly by the people with the right skillset.

By 8:00, Edison is preparing for a customer meeting to provide a status update and discuss upcoming focus and priorities. Edison presents his data analysis, supported by slides that outline the problem, the plan and the expected outcomes. Transparency is crucial. “We explain how we’re attacking the issue at hand, and how we’re going to move forward,” says Edison. “This reinforces the customer’s confidence in the team and the process.”

When a system is down, emotions can run high – and the pressure from the customer is palpable. First-line engineers are on-site for the task, while Edison reviews data traces remotely and coordinates the actions to solve the problem. His methodical, data-driven approach allows him to communicate clearly with the customer, requesting additional access time to the machine only when he can justify the need. Openness paired with evidence-based reasoning is critical in maintaining the customer’s trust, especially under stress.

If the problem can’t be solved by the first-line team, second-line generalists and competency engineering step in. These engineers have expertise in one or more modules of the systems. In the case of Edison, his competency is in positioning, which includes the wafer table, wafer stage and reticle stage.

Remarkably, about 95% of issues are resolved locally. “This really speaks volumes – in the way we organize, plan and build our action plans and solutions,” Edison points out. When local capacity reaches its limits or there’s a gap in the necessary skills, Edison taps into ASML’s regional customer support network.

In addition to managing escalations, Edison is also responsible for scheduled equipment maintenance. Lithography system downtime is very costly for customers because it can reduce fab output. Therefore, maintenance needs to be meticulously planned, reviewed and coordinated to help support maximum uptime. As part of this effort, he uses predictive analytics and dashboards to continuously monitor equipment health, allowing him to anticipate potential problems and maintain peak system performance.

Extreme ultraviolet (EUV) lithography, which uses 13.5 nanometer light, enables the production of the most advanced semiconductor features and plays a critical role in the high-volume manufacturing of today’s leading-edge microchips. However, such light can’t be produced by lamps or even directly by lasers. When ASML started developing EUV lithography about 25 years ago, it became clear that a completely new type of light source was needed.

EUV source

After looking at the alternatives, we chose an architecture based on laser-produced plasma (LPP), which promised a final source that required less downtime for maintenance and offered scalability to higher powers. With LPP, powerful laser pulses are fired at tiny droplets of tin travelling through a (near) vacuum – transforming the droplets into exploding balls of plasma 40 times hotter than the surface of the sun and causing them to emit an intense burst of EUV light.

To create enough EUV light for making microchips, this process is repeated thousands of times per second – 60,000 times per second in our latest commercial sources. The light is collected and channeled into the lithography system’s optics – no small feat, given that EUV light is absorbed by almost all materials. We worked closely with our optical partner ZEISS to develop a system using mirrors made up of more than 100 layers of precision-engineered materials.

Evolving to meet customers production needs

The source in our first prototype EUV lithography system, released in 2010, delivered just one watt of EUV light power – sufficient for customers to begin exploring EUV, but it fell short of the requirements for volume production. 

Reflecting on the early days, Jayson Stewart, now Head of Source Research, recalls: “For a long time, the output wasn’t powerful enough to make EUV a viable investment for customers. But those early wafers that were exposed – albeit very slowly – showed the potential, and we knew EUV lithography could be a key enabler for the industry.”

Creating a powerful and reliable EUV light source is both a physics challenge and an engineering one. We worked with our technology partners to push the forefront of science to improve the plasma explosions and found that flattening the tin droplet before triggering the explosion increased the EUV light output. We also developed a way to prevent feedback between the plasma and laser, enabling more powerful, stable laser pulses.

What steps has ASML taken to deepen collaboration with suppliers?

Building on a foundation of collaboration over the past years, we’ve made significant strides in strengthening our supplier relationships, particularly near our global headquarters in Europe, where our robust network continues to play a pivotal role.

Our headquarters in Veldhoven, the Netherlands, serves as a hub for development, engineering and manufacturing – supporting seamless integration with suppliers on cutting-edge products and technologies that are in ongoing development. This proximity supports real-time collaboration, accelerated innovation cycles and efficient roll-out of new advancements. 

By working hand in hand with our partners, we’ve enhanced our ability to co-develop solutions that address complex challenges in lithography and metrology, maintaining our leadership in high-precision systems. 

What strategies are you implementing to evolve your supply chain and adapt to the diverse needs of global markets?

Looking ahead, we’re committed to evolving our supply chain to support sustainable growth and operational resilience. We’re actively expanding our sourcing footprint to include strategies that promote local repair and reuse, as well as localized sourcing for service operations where it makes strategic sense. This approach can reduce downtime for our customers while also contributing to environmental sustainability by minimizing waste and transportation impacts.

For our mature product lines – where the emphasis shifts from rapid innovation to cost efficiency and unwavering quality – we’re increasingly turning to lower-cost geographies such as Asia. This diversification allows us to optimize production costs without compromising on standards, allowing our established technologies to remain competitive in a global market. 

At ASML, we have evolved how we talk about inclusion over time, to think in terms of ‘Inclusion and Diversity’ instead of ‘Diversity and Inclusion’.  This might seem a minor change – but to me it is hugely significant because while diversity is a fact, inclusion is an act. 
 

Inclusion can make diversity succeed because it contributes to the innovative mindset at ASML. We can be extremely diverse, but if we don’t have an inclusive work environment, people may revert to how they think the majority looks and acts. Whatever that majority is, it’s still mono-dimensional, which is not as rich as a multi-dimensional culture. 

The challenges we have are definitely not mono-dimensional – they require teamwork and different perspectives addressing the same problem, making sure that we flesh out an issue and challenge each other on what’s the right solution.
 

Creating an inclusive environment

We are on a journey to inclusivity – and it’s a long journey. We’ve made good progress, but there is still more to be achieved, which is why we have a number of programs running at ASML. For example, we believe establishing a common language for what inclusion means is a fundamental step towards making improvements. 

We do this through the Ignite Inclusion program, which is an engaging online conversation about what we mean when we say ‘inclusion’. What do we understand about inclusion? Let’s talk to each other and find out. Since its inception in Spring 2024, more than half of our employees have completed the program. We aim for everyone at ASML to do so by mid-2027.
 

As demand for high-performance computing accelerates and reinvigorates Moore’s Law, AI is becoming a growth engine for the semiconductor industry – and AI has long been part of our product leadership at ASML. Our unique partnership with Mistral AI brings together their advanced AI science and our lithography expertise, which we believe will allow us to explore opportunities and deliver customized solutions that can push the boundaries of efficiency and precision.

“We need to have a mindset where we understand what AI can and cannot do to drive our innovation in a responsible way. AI is not a one-size-fits-all solution to everything."

Marco Pieters, EVP and Chief Technology Officer

Improving R&D speed and quality

Our AI strategy in R&D spans software and hardware development and is designed to enable faster innovation and improve quality. In hardware development, we use AI with the aim to shorten development cycles. For example, deep learning surrogates using advanced neural network models can accelerate physical simulations. This has the potential to reduce the simulation time for computational fluid dynamics for a particular component in the EUV source from 24 hours to just seconds, which drastically increases the number of design configurations to be explored for optimization.

In software development, AI tools have shown significant productivity gains and have been rolled out to thousands of developers. Working closely with Mistral AI, we’re adopting customized solutions, trained with ASML source code and documentation. 

AI inside our products and services

Across our holistic lithography portfolio, vast amounts of data are processed to enable continuous feedback loops and integrated optimization. AI is already embedded in our products, particularly in computational lithography and metrology and inspection – it’s used to improve both the speed and accuracy of our optical proximity correction products, for example.
  
Our HMI e-beam metrology and inspection systems also apply machine learning to improve defect detection and guide inspection sampling, while YieldStar optical metrology systems use AI to shorten time-to-recipe to get overlay data from measurements.

Our lithography systems generate an enormous stream of real-time data from over 100,000 actuators and sensors. Harnessing this data is essential for predictive control strategies, for example by using AI to better and faster predict compensation schemes for reticle heating.

In recent years, the most advanced features on cutting-edge chips have become so tiny that they can’t be directly printed by deep ultraviolet (DUV) lithography systems. So, for the critical layers with the smallest features, some of our customers need to make a choice. In many cases, they use DUV systems and a technique called multi-patterning. This involves splitting complex patterns of tiny features into simpler patterns of larger features – printing each one separately using multiple exposures to create the final pattern. Or, they can now use ASML’s extreme ultraviolet (EUV) systems, which use a much shorter wavelength of light and can print the original pattern in just one go.

The consideration of which is the most sustainable route is not a simple one. EUV systems consume more power than DUV systems, mainly because of the required light source and lower transmission of the reflective optics. But, because EUV systems can expose the whole pattern in one go, chips can be produced with fewer process steps. That doesn’t mean just fewer lithography steps, it also applies to etch, film deposition and other steps, which all require energy, chemicals and water. 

So, can we estimate the environmental benefit of EUV from the perspective of the whole chip manufacturing process, given that there are so many steps and variables? 

Well, we can try.

Imec, a leading nanoelectronics R&D hub in Leuven (Belgium), has developed imec.netzero – a virtual model of a high-volume semiconductor fabrication plant (fab). This bottom-up model combines chip manufacturing process data, process flows and fab models with environmental impact models, to calculate the environmental footprint of manufacturing a chip. The model provides insights into key metrics such as the total number of process steps, overall electrical energy consumption (in kWh per wafer) and total wet chemical usage (in liters per wafer). It enables comparisons across different chipmaking technologies introduced between 2013 and 2025.