Finnect AI Insights

Policy · · 5 min read

Why is One-Third of the AI Supply Chain Invisible to Global Policy?

This article reveals why a hidden segment of the AI supply chain is ignored by governments, equipping business leaders to navigate global technology dependencies and build resilience.

200mm Wafer Fertigungslinie

Imagine a vast, intricate network powering the future of technology. Yet, a significant chunk of it operates in the shadows. It's unacknowledged by the very governments trying to regulate it. That's the striking reality revealed by a new University of Cambridge study.

This research found that a large segment of the global AI supply chain receives no attention from national and international policymakers. It's a massive 'zone of invisibility' right in the middle of where AI is made.

This isn't just about a few overlooked parts. The study identified 137 specific 'elements' across 14 supply-chain layers. These are produced in 56 countries by nearly 800 firms. However, policy documents from 43 countries, the EU, and the OECD ignore over 33% of these critical components.

On the surface, it seems like a straightforward oversight. But that's not the interesting part. This isn't just about missing a few details. It points to a fundamental misunderstanding of how AI truly works and where real power lies.

Dr. Siyi Liu, the lead author of the University of Cambridge report, highlights this. She notes that most policy discussions simplify artificial intelligence to just 'compute, data, and models.' This narrow view misses the complex 'physical middle' of production. It's a universal pattern, creating an 'unexpected mismatch' between a nation's stated AI policy and its actual supply chain strengths.

That raises three questions.

Question 1

What Exactly is This 'Invisible' Part of the AI Supply Chain That Governments Are Missing?

When we talk about AI, our minds often jump to large language models like ChatGPT. Or perhaps the massive data centers that train them. However, the University of Cambridge study points to the 'physical middle' of AI production. This is the primary blind spot for policymakers.

This 'physical middle' includes foundational elements. Think wafers (thin slices of semiconductor material), chip design, packaging, networking, and testing. It also covers server manufacture and electronic design software. These are the nuts and bolts. They form the underlying infrastructure that makes AI possible.

Researchers mapped 137 such elements across 14 layers of the AI supply chain. They found these elements are produced or convened in 56 countries and territories. Almost 800 firms are involved. Yet, policy attention tends to concentrate narrowly on models, data centers, funding, and governance. This leaves the physical components largely unaddressed, according to Dr. Siyi Liu.

For context, consider the Japanese firm, Ajinomoto. It holds a 95% market share for a crucial material called build-up film. This film is essential for AI chips. Or take the Dutch firm ASML, a leader in EUV lithography. It relies on vital parts from German and US companies. These specific, often obscure, components are the 'chokepoints.' They often sit outside the perceived AI powerhouses of China and the US. And that brings us to the second question.

Question 2

How Did Policymakers Overlook so Much of It, and Why Does This Invisibility Matter for National Leverage?

Policymakers have largely focused on the 'sexy' parts of AI. Think algorithms, data, and big-name companies. This is like a car manufacturer focusing solely on the engine and sleek exterior. They ignore the complex ecosystem of suppliers for thousands of small, critical components. We are talking about sensors, wiring harnesses, and brake pads.

To understand this oversight, consider an Indian automobile manufacturer. They might proudly announce a new electric vehicle. Their focus is heavily on battery technology and software. But what if they completely overlook the intricate web of small and medium enterprises (SMEs)? These SMEs supply thousands of nuts, bolts, sensors, and wiring harnesses. These components can make up a third of the car's parts. If those suppliers face issues, the entire production halts. It doesn't matter how advanced the battery is.

This 'unexpected mismatch' means many governments are unaware of their own strengths and vulnerabilities. The UK, for instance, is involved in 41 of the 137 identified supply-chain elements. Yet, its AI policies name only three. Japan holds the second-largest number of supply chain positions globally. However, its AI policy documents recognise only three of its 69 elements.

This lack of awareness gives countries like Japan, Germany, Belgium, and South Korea more leverage over global AI production than their governments realise. Verity Harding, Director of the AI and Geopolitics (AIxGEO) project, noted this. She stated, 'no one country can control the AI supply chain alone.' Dr. Liu suggests that understanding these dependencies forms the basis of 'credible bargaining power' in AI governance. And that brings us to the third question.

Question 3

What Does This 'Invisible' AI Supply Chain Mean for Businesses Navigating Global Technology Dependencies?

For businesses, this study screams one thing: resilience. Governments are blind to over a third of the AI supply chain. This means your AI strategy could be built on shaky ground. Supply chain disruptions, geopolitical tensions, or new regulations in these 'invisible' areas could have unforeseen impacts. These impacts could be significant for your operations.

The report suggests that complete national AI supply-chain sovereignty is 'impossible.' This means businesses cannot simply rely on their home country's policy to secure their AI needs. Instead, you must proactively map your own supply chain dependencies. Look beyond the obvious vendors and technologies.

You need to ask: Do we truly understand where our AI chips come from? What materials are critical to their production? Are there chokepoints in countries whose policies we don't track? This isn't just about sourcing. It's about understanding the entire global ecosystem that enables your AI initiatives.

The implications extend to risk management, strategic planning, and investment decisions. Businesses that understand these hidden dependencies will be better positioned. They can mitigate risks, diversify suppliers, and adapt to a global AI landscape. This landscape is far more interconnected and vulnerable than most policymakers currently grasp.

The Bigger Picture

The 'invisible AI supply chain' isn't just a fascinating academic finding. It's a stark reminder that our understanding of critical technologies often lags behind their actual complexity.

For business leaders, this means moving beyond the headlines. Dig deeper into the foundational components that power your AI ambitions. The future of AI isn't just about algorithms. It's about the intricate, often unseen, global industrial web that supports them.

Ignoring this hidden segment could leave your business exposed to risks. Risks that even governments aren't seeing. The real competitive edge will come from those who truly understand what makes their AI tick, from the silicon up.

Quick Answers

What part of the AI supply chain is missing from policy?

Governments are largely overlooking the 'physical middle' of AI production, which includes critical components like semiconductor wafers, chip design, packaging, networking, testing, server manufacture, and electronic design software.

Why do governments overlook parts of the AI supply chain?

Policymakers tend to focus on high-level aspects of AI like models, data, funding, and governance, simplifying the complex reality of AI production and missing the intricate network of physical components and their global suppliers.

What does the 'invisible AI supply chain' mean for businesses?

For businesses, it means potential unseen vulnerabilities and risks in their AI operations. They must proactively map their own AI supply chain dependencies to ensure resilience, as national policies may not cover these critical, overlooked areas.

Sources

Facts are credited to the outlets above. The analysis, calculations and Signal Check are by the Finnect AI Insights desk.