追踪极紫外光刻的出现：识别、保护和促进新兴技术的经验

Lessons for Identifying,Protecting, and Promoting the Next Emerging Technology Author John VerWey Executive Summary This paper presents a case study on the most important technology to have emerged inthe past decade:extremeultraviolet (EUV)lithography.In 2019, when the firstcommercial electronics enabled by EUVwere released,the technologywas hailed as“the machine that saved Moore’s Law.”1All of today’smost advancedartificialintelligence (AI)chips, smartphones, autonomous driving systems, andhigh-performance computerscontain semiconductors fabricated using EUV lithography.TheDutch company ASML has emerged as the sole supplier of EUV machines, winning a30-year race that granted the company a monopoly on selling the tool essential forfabricating leading-edge semiconductors.2 However, while ASML gets well-deserved praise for developing and commercializingEUV, this paper’s focus is on the research community that supported EUV from thebeginning: the academics in Japan, the United States, and Europe;thepublic-privatepartnerships;the conferences;and the industry collaboration that laid the groundworkfor EUV in the 1980s and 1990s.Without this community,“the most technicallyadvanced tool of any kind that’s ever been made”would nothavebeenpossible.3Thispaper traces the academic, government, and industry actors involved in a multi-decademoon-shot project that ultimately saw EUV ascend from a speculative emergingtechnology to themechanismthat makesNvidia's leading-edge AI training chipsandApple’s latest smartphone possible. Careful study of the researchcommunity that supported EUV development isparticularly relevant for policymakers and the semiconductor industry today. EUVresearch began in the 1980s, whenthe U.S. semiconductor industrywas tryingto fendoff ascendent Japanese firms amid significant government intervention on both sides.At the same time, the industry recognized that a new generation of lithographic lightsources would be necessary to fabricate future advanced chips to maintain Moore’slaw. Similar circumstances exist today, with policymakers in the United States, Europe,and Asia engaged in once-in-a-generation efforts to protect and promote theirrespective semiconductor industries, all while ascendentChinesefirms attempt tochallenge industry leaders. Meanwhile,the entire semiconductor industry recognizesaslow-moving existential crisis:rapid advances inAImust be sustained bycorrespondinglyrapid advances in computationalpower.However,the end of Moore’slaw is in sight, and not even EUV can save it.4 Thedevelopment of EUV reflects many of the emerging technology themes observedin previous CSET analyses.5Research collaborationamongacademia, industry, andgovernmenthasoccurred for decades, making progress on associated technologies in Center for Security and Emerging Technology |1 fieldssuch asmaterials science,plasma physics,andchemistry. This progresshas beendocumented in the form of journal article publications, patent filings, and conferenceproceedings. Over time, specific technologieswouldtransition from government labsto the private sector, frequently throughpublic-private partnerships and consortiaformed to address and overcome technical hurdles. Eventually a private firmwouldassess a market opportunity,makeinvestments to increase the technology’s maturity,collaborate with customers, and commercialize the technology. Government supporthas always beenessential at various points along the way. Bibliometric study of the EUV research community during the period when EUV wasstill“emerging”offers important lessons for policymakers interested in identifyingpromisingtechnologies today. It shows how academic research translatesintoscientific advances, the role of government and corporate labs, how internationalresearch collaborations accelerate innovation, the power of public-privatepartnerships, and the need for large amounts of patient private-sectorcapital. Thispaper catalogs this multi-decade path of innovation, identifying inflection points,signals, and sub-innovations along the way. Based on these findings, this paperconcludes by introducing a set of criteriapolicymakers could use when attempting toidentify future emerging technologies. That a Dutch company, ASML, commercialized a technology pioneered in America’snational laboratory ecosystem and largely funded by Intel also has important lessonsfor policymakers interested in protecting and promoting the next emergingtechnology.6International collaboration on emerging technologies is inevitable,andguardrails on this collaboration come with trade-offs. Investments are an importantsource of funding for innovation, yet strategic acquisitions can fundamentally altercompetitive dynamicsinheavilyconsolidated industries. Finally, emerging technologysupply chains evolve and mature in the private sector, whose incentives are profit andreliability, not geostrategic competitiveness.EUV tools are monopolized, but the EUVresearch community and underlying supply