All Insights

Sovereign AI Part 1 - The AI supply chain

Written by Zwan Mahmod | Oct 6, 2026, 4:05:52 PM

Perspective

This article explores the UK's position within the global AI supply chain and the strategic questions this creates. It is intended to stimulate discussion and does not represent a description of services offered by Methods or formal policy advice.

Supply chains have always been strategically important, but the Russia/Ukraine War and COVID-19 pandemic were timely reminders of this salient fact. Disruptions to these chains caused goods shortages, price surges and an increase in unemployment. Alongside pandemics and wars, natural disasters can also affect supply chains. They are inherently vulnerable.

This is part of what makes what we import, and what we produce domestically, so important. In the context of AI and other electronics, what is crucial are semiconductors. The UK imports its semiconductors mostly from East Asia, while it relies on the US for its cloud infrastructure and frontier AI models. In worst-case scenarios, including war, natural disasters and pandemics, the provision of these components needed for AI would be restricted or limited as supply chains are disrupted.

Diversifying the supply of semiconductor imports and expanding the UK's native production are both therefore essential. While Japan, China, Taiwan and the US dominate the current market, and this dominance is set to continue, building up a modest homegrown capability is one course of action worth pursuing.

More importantly, recognising the UK's strengths and leaning into them is even more crucial. This is already part of the government's strategy, as outlined by the UK Semiconductor Strategy report of 2023, which sets out a 20-year vision for the country to:

“Have a world-leading position in the semiconductor technologies of the future by focusing on our strengths in intellectual property and design, compound semiconductors and research and development.”

While traditional semiconductors are typically made of silicon and are used in everyday electronics, compound semiconductors are made from two or more elements. They are particularly suitable for advanced technologies because they allow faster processing and can handle higher voltages of energy, making them well suited to AI models, solar panels and LEDs.

Their importance is recognised and further advanced with the establishment of the Welsh Compound Semiconductor Cluster, a hub of innovation supported by academic expertise from Cardiff University. It receives backing from the Welsh Government, while businesses benefit from research and development and capital grants.

This is a fantastic development which should, where possible, be replicated across the country. This would mean a future in which several hubs coexist and share knowledge. It is particularly important in preventing the possible loss of British semiconductor companies such as Pragmatic Semiconductors and IQE, which have considered relocating to the US or the EU, citing a lack of government support.

High-end technology has a strong historical precedent in the UK, as seen in the example of ARM. ARM developed a chip architecture and licensed it to other companies, which then manufactured the chips themselves. This model has since spread worldwide, and advanced companies increasingly focus on design rather than manufacturing.

The ambition should not therefore be to compete with China, Taiwan or the US in terms of raw production output, which is not materially possible. Instead, the UK should recognise that:

“British comparative advantage is in ideas and specialised hardware, not commodity volume.”

As Christopher Cytera writes, this is what would give Britain significant leverage in the AI supply chain.

While ARM was a British company, in 2016 it became majority-owned by the Japanese company SoftBank. The focus for the UK should be a return to companies like ARM, placing emphasis on semiconductor design.

Supply chains across the world are denser and more vulnerable than ever. It is worth developing our understanding of what the UK's supply chain looks like and asking ourselves: what other possibilities do we have for strengthening our AI supply chain?