The Evolution and Geopolitics of Lithography Technology

Key Insight

The pursuit of Moore's Law, which necessitates ever-smaller transistors, drove a critical challenge in lithography technology during the 1990s. Intel recognized that existing methods were insufficient and in 1992, CEO Andy Grove approved an initial $200 million investment, eventually escalating to billions, into Extreme Ultraviolet (EUV) lithography, despite widespread skepticism that it could be mass-produced. This period was characterized by 'lithography wars' among engineers, debating the merits of various techniques like electron beams, X-rays, and EUV, all aimed at achieving the precision required to etch features as small as a dozen nanometers.

The commercial battle for next-generation lithography tools resulted in industry consolidation due to the enormous development costs. U.S. firms like GCA and Silicon Valley Group (SVG) fell behind, leaving Japan's Canon and Nikon as market leaders. Against this backdrop, ASML, a small Dutch company spun out of Philips in 1984, emerged as a dark horse. ASML's strengths included its unique approach of assembling systems from globally sourced, best-in-class components rather than in-house manufacturing, and its perceived neutrality in the U.S.-Japan trade disputes, making it a trusted alternative for U.S. chipmakers like Micron.

The geopolitical dimension of lithography became evident through the crucial partnership between ASML and TSMC, fostered by Philips' foundational investment in TSMC. While Intel collaborated with U.S. Department of Energy labs to advance EUV research, no American firm could commercialize the technology. With Japanese competitors uninterested or excluded, ASML became the sole viable partner. Despite concerns from the Defense Department and Congress about U.S. reliance on a foreign entity, particularly after ASML acquired SVG in 2001, Washington largely dismissed these geopolitical implications, prioritizing globalization and industry efficiency. This decision led to ASML's effective monopolization of EUV manufacturing, making a single company responsible for a tool fundamental to all future computing.