How ASML Reduced FX Risk by 26% and Improved Customer Support with KNIME

Challenge: Improving Accuracy of Forecasting for Foreign Exchange Hedging and Tool Fragmentation in Customer Support

ASML Treasury Department's USD Forecasting Problem

The treasury department's challenge stemmed from ASML's global supply chain complexity. Raw materials needed to build lithographic materials are purchased in USD, but machines are sold in euros. This created foreign exchange risk that could erode margins. 

Based on 2022 data, 16% of ASML's material intake came from US suppliers, requiring precise USD forecasting for their hedging program. However, the existing process involved multiple departments manually enriching Excel files with exchange rates and currency data. This was problematic because:

• The process was very labor intensive 
• Forecast accuracy was only 70%, meaning significant portions of required USD purchases were missed
• The remaining 30% exposure had to be exchanged at current rates, creating unnecessary FX risk

Customer Support's Tool Fragmentation Challenge

On the technical side, customer support engineers faced a different but equally complex problem. ASML manufactures advanced lithographic systems — complex machines that print circuit patterns on silicon wafers to create semiconductor chips. 

When customers had problems with their products or couldn't achieve the precise pattern needed for advanced chip manufacturing with ASML’s machines, the support team had to diagnose issues across thousands of machine parameters.

Chao Jiang, an IT architect, described the complexity using a coffee machine analogy: 

"At home, you can play with the knobs [on a coffee machine]. There are three. So you can change different values in combination. And eventually, you will somehow get a coffee that you like. But with a [lithographic] machine, we have thousands of knobs."

The engineering challenge was compounded by tool fragmentation. Diagnosing performance issues required engineers to use numerous specialized tools developed in different programming languages. Engineers needed to master multiple toolboxes and their underlying domain knowledge to solve even simple issues like overlay errors, where chip layers printed by the lithographic machines must align within nanometer precision.

ASML were able to use KNIME to solve both of these problems. 

Solution: KNIME as a Unifying Platform

If you copy and paste text, which is made up of several paragraphs, it will transfer to the website as sub-blocks. These can be edited, removed, moved around, etc. by right clicking the block.

ASML initially chose KNIME as their data science platform for two primary reasons:

1. KNIME was able to create transparency in finance applications and; 
2. KNIME could integrate diverse tools in one place to diagnose and resolve customer issues.

Treasury Forecasting Solution using KNIME

For the treasury department, the ability to audit processes with KNIME's visual workflows was crucial. KNIME’s self-documenting visual workflows provided complete traceability by clearly showing each step of the forecasting process and every data operation, ensuring the team was always audit-ready.

ASML’s solution architecture for KNIME included several key components:

• Data integration: Direct connections to SAP and HANA Cloud systems eliminated the need for manual data extraction, providing smooth access to production data without safety concerns.
• Configurable workflows: The team built modular KNIME workflows with configuration nodes that could be easily adjusted for different dates, files, accounts, and cost centers.
• Hybrid modeling approach: The solution combined KNIME's built-in nodes with custom Python code. Initial development used built-in nodes for feasibility checks, then custom models for monthly retraining with changing configurations.

"It is like having the best of two worlds. I like to do things in Python in my own way, and also we have KNIME nodes to build the workflow."  Zeinab Bakhtiarinoodeh, Senior Data Scientist in the finance team at  ASML.

ASML followed a structured approach to solution development, emphasizing rapid iteration and business validation. The treasury team used a three-phase approach:

1. Proof of Concept: Three months of development and validation
2. Pilot Phase: Three months of shadow running alongside existing processes
3. Production: Full deployment with ongoing monitoring

This approach required a platform that enabled fast development where they could develop fast and fail fast and KNIME fit the bill perfectly.  

The team developed innovative approaches to documentation using KNIME's workflow capabilities. Zeinab noted:

"Usually in the beginning of the project, I use the workflow executor summary. And then we extract this. And now, manually, I feed it to the Copilot to create documentation and auditing for the workflows."

Customer Support Integration Platform using KNIME

ASML built custom connectors in KNIME, allowing engineers to integrate their existing analysis tools into unified workflows for customer issue resolution. 

Hans Zeng, a senior application engineer, explained how they approached this: they created connections which enable engineers to "put our toolbox and script, which we use to analyze product performance, directly into a node.”

This integration made it possible for engineers without deep expertise in every tool or programming language to perform complex analyses efficiently to diagnose and resolve customer issues, saving time, effort, and streamlining everything into one tool –– KNIME. 

When a workflow proved effective in solving a real customer issue, it was documented and made reusable with KNIME Business Hub. Other engineers could then reference these workflows, learn from previous solutions, and quickly adapt them to new problems. 

This approach not only sped up issue resolution and therefore increased customer satisfaction with support, but also fostered continuous learning and created a knowledge-sharing system across the engineering teams.

Results

Reduced FX Risk, Faster Customer Support, and Efficiency Improvements

The treasury forecasting solution helped ASML achieve the following results:

• Accuracy increase: From 70% to 96% forecast accuracy
• Process efficiency: Reduced from multiple departments to only one FTE working on monthly checks
• Business impact: Direct impact on gross margin through minimized FX exposure
• Recognition: The solution was nominated for three prizes and won all three for being innovative, simple, and using open source tools and technology

Customer Support Knowledge Platform

The engineering platform enabled better collaboration and knowledge transfer:

• Engineers could use complex analysis tools without deep technical knowledge of each toolbox, while reusable workflows captured institutional knowledge for future problem-solving
• The platform served as a learning system where engineers could understand how issues were solved previously and adapt solutions for new use cases

Future Applications

ASML identified additional opportunities for KNIME usage, particularly in product development:

• The team recognized potential for using KNIME to connect internal toolboxes and generate results for customer demonstrations, enabling quick validation of whether proposed software features would be interesting to customers
• This rapid prototyping capability could significantly reduce development cycles by enabling quick customer validation before full product development