JC Xue, Portfolio Manager at Foord Asset Management Singapore
At first glance, AI and deglobalisation seem to belong to different stories. AI speaks to technological progress, innovation and the next phase of productivity. Deglobalisation suggests fragmentation, higher costs and the unwinding of decades of economic integration. That said, the two forces are fast becoming closely connected.
For much of the past three decades, the global economy was organised around efficiency. Companies produced goods wherever labour was cheapest and supply chains stretched across continents. Digital infrastructure could be located wherever costs, tax rates and regulation were most attractive. The aim was to minimise costs. Consumers benefited from lower prices, inflation was contained and many emerging markets prospered as they became part of global production networks.
That model is now changing. Governments are more focused on the control of key technologies than on the price of goods and services. Data, computing power, semiconductor production and cloud infrastructure are increasingly treated as strategic national assets. Europe talks about digital sovereignty. China emphasises technological self-sufficiency. The US frames the issue in terms of national security. The language differs, but the objective is similar: countries want more control over the technologies on which their economies depend.
The result is a race to build domestic technological infrastructure. Countries want local semiconductor capacity, domestic data centres, secure cloud networks and national AI capabilities. This has created a powerful investment boom, especially in the US. Spending on data centres, chips, power infrastructure and computing capacity has helped offset some of the drag from tariffs and trade uncertainty. Without this AI-related capital spending, recent US growth would probably have looked much weaker.
The irony is that the AI revolution is not reinforcing the highly efficient global system that emerged after the Cold War. Rather, in many ways it is encouraging countries to duplicate capacity. Semiconductor plants are being built in more places. Data centres are being located closer to end-users and regulators. Supply chains are being redesigned not only for cost, but also for security and political acceptability.
This provides resilience, but it is not free. Building parallel systems is less efficient than relying on a single integrated global network. Domestic production often costs more than offshore production. Duplicate supply chains require more capital. Power grids must be expanded. Skilled labour must be found. Over time, this points to structurally higher costs and more persistent inflationary pressure than the world became used to during the era of hyper-globalisation.
History also suggests that transformative technologies can be both economically important and financially dangerous. Railways, electrification, telecommunications and the internet all changed the world. They also produced periods of overinvestment, excess capacity and poor returns for many investors. While the infrastructure was useful, the investment outcomes were mixed.
AI may follow the same pattern. It could raise productivity, reshape industries and create large new companies. Yet the rush to control the technology may also lead to duplicated investment and lower returns on capital. The political logic is clear, but the investment case is more complicated.
The age of hyper-globalisation was built on efficiency. The age of artificial intelligence may be built on resilience. That may make the global economy safer in some respects. It is unlikely to make it cheaper.
ENDS
