ASML’s $9B R&D Bet: How Installation Complexity Creates Lock-In

More than $9 billion in research investment needs to be recouped, spread across relatively few machines sold each year. Unlike software where R&D costs spread across millions of users, ASML spreads comparable R&D across perhaps 50-100 machines annually at the high end.

R&D Amortization Economics

ASML’s pricing reflects a unique R&D amortization challenge. Each machine must carry an enormous share of development costs. The EUV program took over 20 years from concept to high-volume manufacturing.

The Japanese consortium abandoned the effort in the early 2000s because it was deemed too risky. No one knew how long it would take to be successful. ASML persisted, backed by customer investments from Intel, TSMC, and Samsung who needed the technology to exist.

This co-investment model de-risked the R&D while locking in future customers. Multi-decade R&D cycles and $9B+ investments are amortized over dozens of machines. High ASPs are structurally required, not opportunistic.

Installation Economics

The installation process of ASML’s High-NA Twinscan EXE 150,000-kilogram system required 250 crates, 250 engineers, and six months to complete. This creates several economic effects:

High switching costs: Once a fab is built around ASML equipment, the integration runs too deep to switch. Manufacturing processes are optimized for specific machine characteristics.

Service revenue stickiness: Beyond the $300M price tag, customers also pay $10-15M annually for maintenance, upgrades, and software optimization. Each machine generates recurring revenue — as explored in the shift from SaaS to agentic service models — for decades.

Capacity constraints as pricing power: With 6-month installations requiring 250 engineers, ASML’s delivery capacity is inherently limited. Customers compete for machine slots years in advance.

This is part of a comprehensive analysis. Read the full analysis on The Business Engineer.