The road to TSMC Part 1 of 2: the DARPA challenge behind TSMC's success

TSMC's sub-28nm moat is rooted in US public research funding. Calvin Hu's nanowire transistor work at Berkeley—born from a DARPA 25nm breakthrough mandate—directly became the architectural backbone of TSMC's 20nm FinFET mass production. The first push behind the world's most advanced process was taxpayer-funded basic science, not corporate R&D. This lineage locks upstream supply chains (high-k dielectrics, EUV photoresist) and downstream roadmaps (Apple, NVIDIA). Re-examining IP provenance would fundamentally rewrite the technological-sovereignty narrative. The real risk sits in compliance. DARPA-funded patents carry government retention clauses; if Washington invokes these in export-control reviews, TSMC's advanced-node licenses in Taiwan, China and Arizona face an unprecedented legal gray zone. CHIPS Act conditions already signal a provenance-to-access binding mechanism. Samsung's 3nm GAA and Intel's 18A acceleration partly replicate the public-funding-leverages-frontier playbook. Korea's semiconductor strategy and the EU Chips Act are nationalized versions of the same logic. 12-24 month outlook: IP provenance audits will become routine in US export reviews; the Arizona fab becomes the physical anchor of sovereignty contestation; GAA-node competition forces rivals into deeper government-academia alliances. Single-source dependency is structurally irreversible.