Every government now has an AI strategy and a budget to match. What separates the countries that turn those strategies into results from the ones that do not is rarely the size of the budget. It is whether investors, citizens, and institutions trust that the plan will survive contact with the next election cycle.
This is the argument I have made in recent columns for La Razón (Peru) and La República (Colombia), drawing on what South Korea's development experience over the past several decades actually teaches, as distinct from the version of that story usually told abroad.
Resources Were Never the Explanation
South Korea's growth is often explained as a story of hard work, export discipline, or simply good timing. Those factors mattered, but they are not what separates Korea from other countries that tried similar strategies and stalled. The real differentiator was execution capacity: the state's ability to set a development plan and actually carry it through multiple political administrations without the plan being gutted, redirected, or abandoned every time power changed hands.
That is a much harder thing to build than a budget line. It requires strategic goals that sit above electoral cycles, agreed on across the executive, the legislature, and local government, so that a semiconductor cluster or an infrastructure buildout begun under one administration is still being funded and defended five governments later.
What an AI Economy Transition Actually Requires From a State
Korea's current push toward an AI-driven economy is organized around a small number of large, capital-intensive undertakings: semiconductor manufacturing capacity, data center buildout, and advanced robotics and physical AI. None of these succeed on private investment alone. They require a state that does the unglamorous, structural work: securing reliable energy supply at industrial scale, building the physical infrastructure investors cannot build for themselves, simplifying administrative approval processes that would otherwise stall projects for years, and developing a workforce pipeline matched to where the technology is actually going.
The state's role here is not to replace private enterprise or pick winners inside it. It is to make the investment case for private capital viable in the first place, then get out of the way. Governments that try to do both, direct the technology and also build the enabling conditions, tend to do neither well.
Political Cohesion Is the Scarce Resource
The distinction that matters is not left versus right, or interventionist versus market-oriented. It is whether a country's political system can treat certain growth engines as above ordinary political contestation. Reversing a semiconductor investment strategy every time administrations change is not caution. It is the single most reliable way to make sure the investment never pays off, because capital does not commit to a plan it expects to be rewritten in four years.
Why This Matters Beyond Korea
I have written specifically about Peru's situation because the opportunity there is real and time-limited. A new constitutional presidency creates a genuine window to modernize institutions that have been overdue for reform for years: cadastral and land registry systems, digitalized public administration, and a clearer path for informal economic activity to formalize. None of this is about copying Korea's specific industrial choices. It is about recognizing that the sequencing Korea used, building institutional execution capacity before the technology arrives, rather than after, is transferable to any country trying to position itself for an AI-driven global economy.
Trust as Infrastructure
The conclusion I keep returning to is that trust functions as infrastructure in a very literal sense. Roads and power grids are visible. Institutional trust, the confidence that a regulatory framework, an investment incentive, or a development plan will still exist in its current form five or ten years from now, is invisible but does the same load-bearing work. Countries that build it attract long-horizon capital. Countries that do not, no matter how large their AI budget announcement, end up competing for the same short-horizon capital that leaves as quickly as it arrived.
Technology is available to everyone now, largely on the same terms. What is not available to everyone, and what actually decides the outcome, is the institutional discipline to make a long-term plan credible across a decade of political change. That is the competitiveness question every government claiming an AI strategy should be asking itself first.
