Governments overlook critical links in AI supply chains
Just over 33% of the elements underpinning global AI production are absent from national and international policy documents, according to University of Cambridge research. An analysis of 117 AI-related policy documents from 43 countries, the EU and OECD found attention concentrated on models, datacentres, funding and governance, while essential manufacturing technologies were overlooked.
Researchers mapped 14 supply-chain layers containing 137 elements, exposing dependencies that extend far beyond the US and China. Blind spots include chip-design software, semiconductor chemicals, testing and advanced packaging. Synopsys, Cadence and Siemens EDA dominate software essential to designing AI chips, yet receive almost no policy attention. Japan’s Ajinomoto controls more than 95% of the global supply of build-up film, an insulating material needed for advanced AI chip manufacturing.
The mismatch also obscures national strengths. The UK participates in 41 of the 137 identified elements, but its policies name just three, overlooking capabilities in chip design and industrial gases. Researchers warn that disruption to upstream components could cascade into AI shortages and wider economic instability. They describe complete national AI sovereignty as impossible and have developed a toolkit to help governments assess dependencies, diversify suppliers and identify where domestic capacity could be built.