What's Happening?
A new report from Cambridge researchers reveals that as much as a third of the global Artificial Intelligence (AI) supply chain remains 'invisible' to national and international policymakers. The study, conducted by the AI & Geopolitics Project (AIxGEO)
at Cambridge's Bennett School of Public Policy, analyzed 117 AI-related policy documents from 43 countries and territories, the EU, and OECD. It found that policies primarily focus on AI models, data centers, funding, and governance, while overlooking crucial 'upstream' components such as factories, chemicals, chip-design software, and packaging technologies. The research identified vulnerabilities across 14 supply-chain layers, with 137 elements produced or convened in over 50 countries. Many critical 'chokepoints' are located outside major AI powerhouses like the U.S. and China. For instance, one Japanese firm holds a 95% market share for a material essential for AI chips, and the Dutch firm dominating EUV lithography relies on parts from German and U.S. companies. This oversight means governments are unaware of significant strengths and vulnerabilities within their own AI supply chain positions, making total 'AI sovereignty' an impossible goal for any single nation.
Why It's Important?
This 'zone of invisibility' in the AI supply chain poses a significant risk to U.S. technological leadership, economic stability, and national security. A disruption to any of these overlooked 'upstream' components could trigger a domino effect, leading to AI supply shortages that destabilize stock markets, investment, and productivity globally. For the U.S., which is central to the AI story, this means that its AI ambitions could be undermined by dependencies on materials or technologies controlled by other nations, some of which may not even realize their own leverage. The lack of comprehensive policy attention to these critical elements prevents effective risk mitigation, strategic planning, and the development of resilient AI ecosystems. It also means that U.S. policymakers might be misallocating resources or failing to capitalize on existing strengths within the broader global AI production network, potentially hindering innovation and competitiveness.
What's Next?
The Cambridge researchers have developed a toolkit to help governments audit their place in the AI supply chain, which could be adopted by U.S. policymakers. This tool would enable them to understand where and when to diversify across AI element providers, identify critical dependencies, or decide when to build national capacity. The findings suggest an urgent need for U.S. policy to broaden its focus beyond finished AI products and models to encompass the entire ecosystem, including raw materials, specialized manufacturing, and software tools. This could lead to increased investment in domestic production of critical components, strategic international partnerships to secure supply, and more comprehensive risk assessments. Furthermore, the U.S. may need to engage in diplomatic efforts to coordinate with allies and 'middle powers' that hold significant, yet often unrecognized, leverage in the global AI supply chain, fostering collective agency and more robust AI governance.
Beyond the Headlines
The 'invisible' AI supply chain raises profound questions about the nature of global power and interdependence in the age of advanced technology. The study reveals that true technological sovereignty is an illusion, as even the most powerful nations are deeply reliant on a complex web of international suppliers for foundational components. This interdependence could either foster greater international cooperation or become a new battleground for geopolitical competition, as nations vie for control over critical chokepoints. Ethically, the lack of transparency in the AI supply chain could also obscure human rights issues, environmental impacts, or labor practices associated with the extraction of raw materials or manufacturing processes. Understanding these hidden layers is not just an economic or security imperative, but also a moral one, demanding a more holistic and responsible approach to AI development and governance on a global scale.













