Back to News
PolicyAI Understanding briefing

Cambridge study finds a third of global AI supply chain hidden from policymakers

A University of Cambridge report reveals that roughly 33% of AI‑related supply‑chain elements are absent from national and international policy documents, exposing a “zone of invisibility” that could threaten AI‑related economic stability.

4 min readRead the linked source
Source-provided image accompanying Cambridge study finds a third of global AI supply chain hidden from policymakers
Source referenceSource recorded
Publisher
techxplore.com
Source type
Linked source — primary-source status has not been established.
ContextUnderstand this in 60 seconds

Key terms

AI Governance
Policies, standards, and oversight mechanisms that guide how AI is developed and used in society.
Compute
The processing resources required to train and run models, often measured in FLOPS or GPU hours.
Test yourselfAI Ethics Quiz

What happened

A research team at the University of Cambridge’s AI & Geopolitics Project (AIxGEO) published a comprehensive mapping of the global AI supply chain, identifying 14 distinct layers and 137 specific elements spanning raw minerals, gases, wafers, chips, servers, and models. The study found that about one‑third of these elements are not mentioned in any of the 117 AI‑related policy documents from 43 countries, the EU, and the OECD. The analysis highlighted critical chokepoints outside the traditional AI powerhouses of the United States and China, including a Japanese firm that controls 95% of a material needed for AI chips and the Dutch lithography leader ASML, which depends on parts from Germany and the U.S. The report also produced a toolkit for governments to audit their positions in the supply chain and suggested that true “AI sovereignty” is unattainable without coordinated international effort.

The AIxGEO team mapped 14 layers of the AI supply chain, from raw minerals and ultra‑pure gases to wafers, chips, servers, and trained models, plus cross‑cutting layers like governance standards and shipping routes. Within these layers they identified 137 distinct elements (e.g., gallium, hyperscale cloud platforms) and 351 sub‑elements, involving nearly 800 firms across 56 countries and territories.

Analysis of 117 AI‑related policy documents from 43 nations, the EU, and the OECD showed that only about 33% of the identified elements are referenced. The “invisibility” is concentrated in the physical middle of the chain—chip design, packaging, testing, and supporting hardware/software such as EDA tools—areas that receive virtually no policy attention despite their critical role.

Key chokepoints highlighted include a Japanese company (Ajinomoto) that supplies over 95% of the global build‑up film used in advanced AI chip manufacturing, and the dominance of three U.S. firms (Synopsys, Cadence, Siemens EDA) in EDA software. The Dutch lithography leader ASML, essential for producing AI chips, relies on components from German and U.S. suppliers, illustrating the interdependence beyond China and the U.S.

The report notes that the United Kingdom appears in 41 of the 137 elements but its policy documents mention only three, ignoring strengths in chip design (Arm) and industrial gases (Linde). Japan, despite holding the second‑largest number of supply‑chain positions, also lists only three elements in its AI policy.

To address these gaps, the researchers created a toolkit for governments to audit their supply‑chain positions, assess dependencies, and decide where to diversify, build domestic capacity, or exit reliance on external suppliers.

Source details: techxplore.com

Why it matters

The findings underscore a major blind spot in : policymakers worldwide focus on high‑level inputs such as , data, and models while overlooking the physical and engineering layers that actually enable AI hardware. This mismatch creates systemic risk—disruptions to upstream components like semiconductor gases, packaging, or electronic design automation (EDA) software could cascade into AI‑related supply shortages, affecting markets, investment, and productivity. By exposing which nations and firms hold leverage in these hidden layers, the study provides a factual basis for more nuanced diplomatic negotiations and strategic diversification, helping governments avoid over‑reliance on a few suppliers and better align policy with the full complexity of AI production.

Policymakers’ narrow focus on , data, and models overlooks the upstream hardware and material layers that are essential for AI system performance. This creates a false sense of security and hampers effective risk management.

Supply‑chain disruptions—whether from geopolitical tensions, natural disasters, or export‑control measures—could quickly translate into AI hardware shortages, affecting everything from cloud services to autonomous systems, and potentially destabilizing financial markets tied to AI investments.

Understanding the true distribution of leverage reveals that middle‑power nations (e.g., Japan, Germany, Belgium, South Korea) possess significant bargaining power that is currently under‑utilized in international discussions.

The study’s toolkit offers a practical method for governments to align policy with reality, enabling more targeted diplomatic engagement and strategic investment in domestic capabilities where they matter most.

By exposing the “zone of invisibility,” the report challenges the narrative that AI sovereignty can be achieved through isolated national efforts, emphasizing the need for coordinated, multilateral approaches.

Interactive Mechanism

Interactive Mechanism: How It Actually Works

Explore the underlying technology behind this development interactively.

Agent Lifecycle Stage:
1
User Intent & Planning: "Audit customer refund request #4092 and settle payment."
2
Tool Calling: Emits structured JSON call crm_get_transaction(id='4092').
3
Guardrail & Verification:Paused: High-value action requires human operator sign-off.
4
Final Settlement: Refund recorded, email receipt dispatched, and audit log stored.
Core takeaway: An AI agent is not just a language model—it is a closed loop of planning, tool invocation, and environment feedback. Production systems require self-healing retries and strict human approval guardrails.
Interactive Concept Check+10 Points
AI Ethics Quiz

Why can ethical evaluation not be reduced to one model score?

What to watch next

Future policy revisions that incorporate the study’s supply‑chain layers, potential alliances among middle‑power nations to collectively manage chokepoints, and industry responses such as diversification of chip‑design software or advanced packaging sources. Watch for any governmental adoption of the Cambridge toolkit and for shifts in trade or export‑control discussions that address the newly identified “invisible” elements.

Adoption of the Cambridge toolkit by national governments or intergovernmental bodies, which would signal a shift toward more comprehensive AI supply‑chain policy.

Legislative or regulatory initiatives that expand policy coverage to include chip‑design software, advanced packaging, and critical raw materials.

Formation of strategic alliances among countries identified as having hidden leverage (e.g., Japan, Germany, Belgium) to collectively manage chokepoints and negotiate with dominant suppliers.

Industry moves to diversify sources for EDA tools, packaging services, and specialty gases, potentially reducing concentration risk.

Any changes in export‑control regimes or trade agreements that specifically address the newly highlighted supply‑chain elements.

Related guides & quizzes

Found this useful?