Frontiers of Trade: Testing future competitiveness with the Flint FAST model

This is the fifth piece in Flint’s Frontiers of Trade series, which looks beyond today’s tariffs and geopolitical tensions to the questions businesses will need to confront over the next decade. Previous instalments examined resources mined in space, autonomous AI agents, trusted information at the border, and the national-security treatment of biological capability. This piece turns to a different frontier: what happens when the sources of comparative advantage themselves become less stable, and what does that mean for long-term business strategy?

A factory, data centre or operating model may last for decades; the economics that justified its location can change within years. Imagine a business deciding where to build its next production facility. Businesses make these decisions by comparing factors such as labour costs, skills, energy, infrastructure and access to suppliers. But increasingly, the challenge is not only that the value of those inputs can change. The weight attached to them can change too.

Automation can reduce the importance of labour costs while increasing the value of engineering skills, capital and energy. Surging demand for compute can make grid access more important than construction or hardware costs. Policy can create new advantages through subsidies or remove them through export controls. And physical constraints, from water stress to extreme weather, can make an otherwise competitive location less viable.

None of this is hypothetical: subsidy packages worth tens of billions have been introduced on both sides of the Atlantic; EU industrial electricity prices were roughly twice US levels in 2025, and higher still in the UK; drought in Taiwan has forced chipmakers to truck in water to keep fabs running; and AI-driven demand for compute is making access to power a constraint on where new capacity can be built.

Semiconductors show how these forces can interact. Subsidies introduced on security and supply-chain grounds supported new capacity; unforeseen AI demand then increased the strategic value of advanced chips; the attractiveness of potential locations became increasingly dependent on power availability; and water stress and grid constraints affected where capacity could actually be built.

The result is a harder problem for long-term business strategy. It is not enough to ask which locations are competitive today, or even how today’s costs might change. Businesses increasingly need to ask which sources of competitiveness will matter tomorrow, how much weight they will carry, and what that means for the next investment or sourcing decision.

The shifting weights of advantage

Comparative advantage is about relative, not absolute, cost: an economy gains from specialising where its opportunity cost is lowest. Those relative costs reflect labour, capital, skills, technology, productivity, natural resources, infrastructure, energy and geography.

For businesses, however, the relevant question is more concrete: how do changes in these factors alter the relative competitiveness of the locations in which they could invest, produce or source? A shock need not fundamentally alter an economy’s comparative advantage to change a commercial decision. A sufficiently large movement in absolute costs, productivity or feasibility can still make one location more attractive than another.

Many of the underlying factors have historically changed slowly enough for businesses to treat existing patterns of specialisation as relatively durable. That assumption is becoming harder to sustain.

Technology has always affected trade. What AI and automation can change is not just the cost of production, but the relative importance of different sources of advantage.

In manufacturing, for example, automation can reduce the weight of labour cost while increasing the value of engineering skills, energy security, capital and access to advanced technology. In services, AI may change which tasks can be delivered remotely, how much labour they require, and which skills command a premium, making compute, connectivity, data access and regulation increasingly important alongside wages and human capital. For tradable services, the equivalent constraint may be the resilience of power, telecoms and digital infrastructure.

Mapping how advantage moves

Measures such as revealed comparative advantage show where economies are relatively specialised in exports today. They are less useful for asking where advantage could emerge tomorrow, especially when the drivers of advantage are shifting quickly.

A forward-looking assessment therefore needs to start with a transparent baseline for relative competitiveness, then ask two different questions: what happens when the value of an existing input changes, and what happens when the importance of that input changes?

In modelling terms, this means introducing shocks to economically meaningful variables and quantifying how relative positions respond. Crucially, the weights should differ by value-chain stage: semiconductor design, fabrication and assembly, for example, reward very different combinations of skills, capital, labour, energy and technology. The results must also be constrained by what is physically and technologically feasible.

The same analytical approach can be applied to internationally traded services, although the relevant variables will differ.

Semiconductors: modelling manufactured advantage

Semiconductors are a useful test case because different stages of the same value chain depend on radically different sources of advantage. Design, fabrication, assembly and testing have clustered in different locations according to the skills, capital, supplier networks and labour required for each. Those clusters remain powerful, but the economics of the next investment will also be shaped by policy support, energy and water, physical resilience and access to essential technology.

Several factors can move simultaneously. Subsidies might lower the effective cost of a new plant; energy and water constraints can raise operating risk; export controls could limit access to equipment or designs; and AI-related demand can increase the strategic value of advanced capacity. A location can therefore improve on one dimension while becoming less viable on another. That is why a single headline measure – whether wages, subsidy value or current exports – cannot describe future competitiveness on its own.

The commercial consequence is not only where capacity is built, but which suppliers can serve which markets, at what landed cost and with what effect on trade flows.

The Flint Future Advantage Stress Test (FAST) provides a way of modelling these shifts. In this stylised application, we compare six leading semiconductor economies across 11 sources of competitiveness. Each is translated into a common cost-equivalent measure, and weighted according to its importance to design, fabrication or assembly. For each stage, the most competitive location is used as the benchmark, allowing us to measure how far other locations sit behind it and how that gap changes under different scenarios. The model is illustrative rather than predictive: its purpose is to test how shocks to different sources of advantage transmit into relative competitiveness.

We then apply three hypothetical shocks:

• Robot abundance: General-purpose industrial robots become cheap, reliable and widely available. Direct labour requirements fall, with capital, electricity, digital infrastructure and robotics expertise carrying more weight. Does this narrow the advantage of established low-cost assembly centres enough to warrant changing supplier choice, capacity commitments or assembly trade?
• Compute rationing: Governments introduce licensing and domestic-priority rules for advanced compute, AI models and semiconductor design tools. Restricted access lowers design productivity and raises effective costs. At the leading edge though, the loss of an essential tool stops being a cost penalty and becomes a feasibility constraint: a location can look competitive on every other measure and still be ruled out. Does that change where a business locates design and engineering capability, or who it can credibly partner with?

• Trusted product passports: Major markets begin to require machine-readable proof of component origin, regulatory compliance, carbon content and supply-chain integrity. Market access, regulation, digital infrastructure and usable data then become direct sources of trade advantage in their own right: compatible suppliers face faster clearance and lower verification costs, while products that cannot be authenticated face delay or exclusion. Does that shift sourcing towards suppliers whose data systems are already compliant, even where they are not the cheapest?
  
  

Three findings stand out:

• First, apparently transformative technologies need not transform relative competitiveness. Robot abundance lowers assembly costs almost everywhere, but because competitors benefit simultaneously, relative positions move much less than the scale of the technological change might suggest. The United States gains most, by 1.2 index points, while Korea narrowly overtakes Taiwan.
• Second, access to an essential input can matter more than underlying cost competitiveness. Under compute rationing, China’s design position deteriorates sharply, while the United States improves. At the technological frontier, restricting access can ultimately turn a cost disadvantage into a feasibility constraint.
• Third, competitiveness can change without a location itself becoming more or less productive. Trusted product passports penalise locations whose inputs cannot readily be verified. Germany and the United States therefore improve relative to China despite no change in their underlying production costs. A regulatory requirement has effectively created a new source of trade advantage.

The next investment is what matters

None of this implies that existing semiconductor clusters will suddenly relocate. Sunk capital, supplier networks, skills and agglomeration effects make established production sticky.

The economically relevant margin is instead where the next contract, source of supply or increment of investment goes. A relatively small change in competitiveness can alter a shortlist, change the sourcing mix for future capacity, increase the resilience value of a second location or strengthen a host government’s bargaining position. That is why movements that leave today’s overall ranking intact can still matter commercially.

What businesses need to model

The question for businesses is not simply what a government has announced or which technology is emerging. It is how the change transmits into the economics of their actual footprint.

Applied to a company’s sites, suppliers and markets, the framework can test how a policy, technology or physical shock changes costs, productivity and feasibility; whether it changes the relative importance of different sources of competitiveness; and what that means for the next sourcing, capacity or investment decision.

It can also distinguish between two very different outcomes: a movement in relative competitiveness and a threshold event that rules an option in or out altogether. And because competitiveness is relative, the analysis needs to consider not only what happens to the business’s existing locations, but how competitors, suppliers and customers are affected by the same shock.

That matters because the determinant of advantage in ten years may be something that barely features in today’s model. The answer is not to assume that today’s conditions will persist, but to understand how different shocks would transmit through the business and where they would change the commercial outcome.

If you are assessing where to invest, source or build resilience, Flint’s trade and economic modelling experts can apply this approach to your actual footprint – quantifying how policy, technology and physical shocks could change the relative attractiveness of locations, suppliers and markets, and what that means for the decisions you need to make.

This blog was written by Zoe Forbes (Associate Director), Surjinder Johal (Director) and Andy King (Specialist Partner).

Flint's experienced economic modelling team applies rigorous analytical methods to help clients understand complex markets, assess policy options, and quantify the impacts of strategic, regulatory and trade decisions. Our economists design models tailored to client needs, from scenario and risk analysis to full macroeconomic and sectoral impact assessments. The analysis in this blog is one example: quantifying how policy, technology and physical shocks reshape competitiveness, and what that means for investment, sourcing and trade.

Enjoyed this blog? Sign up for our Trade Monthly newsletter for analysis of the political, policy and regulatory developments shaping UK, EU and global trade.