AI policy tends to focus on models, data centres and chips, but a Cambridge study says more than a third of the supply chain’s elements are absent from the policy documents it analysed. That blind spot reaches into the hardware, materials and specialist tools that make AI possible, and gives countries beyond the US and China more leverage than they may realise.
Watch Desk analysis
What happened
Researchers from the University of Cambridge’s AI and Geopolitics project mapped 14 layers of the global AI supply chain, identifying 137 elements across more than 50 countries. They compared that map with 117 AI-related policy documents from 43 countries, as well as the EU and OECD. Their study says just over 33 per cent of the elements were not mentioned in those documents, with much of the omission in the physical middle of production: chip design, packaging, testing and the equipment and software behind computing.
The Cambridge study points to concrete dependencies. Japan’s Ajinomoto supplies more than 95 per cent of the global market for a material used in advanced AI chip manufacturing, according to the research. The UK, meanwhile, is involved in 41 of the 137 elements identified, but its policy documents mention only three.
Key findings
- Policy misses more than a third of mapped elements
The researchers say 33 per cent of the supply chain elements they identified are not mentioned in the policy documents studied. - The gaps are in the machinery behind the headline chips
Chip design, advanced packaging and testing, as well as electronic design software, receive little policy attention. - Supply-chain influence is widely distributed
The study identifies 137 elements in more than 50 countries, with important dependencies beyond the US and China. - National policy can miss home-grown leverage
The UK appears in 41 elements but names just three in its AI policies; Japan has the second-largest number of supply-chain positions, the study says.
Why it matters
A disruption to a material, design tool or packaging operation can ripple through production even when access to leading chips and models looks secure. The researchers warn that these dependencies could contribute to AI supply shortages, with consequences for computing capacity, investment and productivity.
The findings also complicate the idea that any one country can achieve complete AI supply-chain independence. If governments want credible options to diversify suppliers or build capacity, they need to know where their dependencies and strengths actually sit. The study says its team has created a toolkit to help policymakers audit that position.
Our read
This is a useful correction to the familiar AI map, where the spotlight tends to land on a handful of labs and chipmakers. The less glamorous links in the chain can matter just as much, which is inconvenient for policy plans drawn with a very short list of boxes.
The research gives policymakers a reason to look beyond finished chips and data centres. Its map is a starting point for that work, not proof that every overlooked element is equally vulnerable or strategically important.
What to watch
- Whether governments use the researchers’ toolkit to map their dependencies and strengths.
- Whether AI policy starts addressing chip design, packaging, testing and the software used to design chips.
- Whether countries revise their industrial or supply-chain strategies in response to the findings.
Discussion spark: Should governments prioritise mapping and diversifying the less visible links in AI supply chains, or risk making policy more complex than the dependencies warrant?
Sources and evidence
- Source update (18 November 2026, 00:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.