The Window Is Open. What Do You Build While It Is?

AI does not permanently change the relationship between population and output. It creates a window. A period where small, fast-moving countries can produce at a scale that their population size would normally prevent. What you build during that window matters more than the window itself.

I have been thinking about this in terms of a simple economic tension. Before AI, a country of 10 million could not match the absolute economic output of a country of 100 million, even if its per-capita productivity was higher. Population sets a ceiling on total output. That has been the ceiling.

AI changes that math. But only while adoption is uneven. The moment the large-population countries embed AI at the same depth, population reasserts itself. The window is not permanent. It is structural to the adoption curve.

The British parallel

I keep returning to one historical case. In the early 19th century, Britain had around 16 million people. France had 30 million. China and India each had populations that dwarfed both. By the standards of that era, Britain should have been an economic footnote.

Instead, it became the center of the world economy for roughly a century. By 1870, British GDP per person was about 70% higher than France or Germany. One third of the British workforce was in manufacturing. Britain had found a technology multiplier: the industrial revolution gave a small population the productive capacity of a much larger one.

Then the window closed. By 1900, the United States and Germany had industrialized on a comparable scale. They had larger populations. Population reasserted itself. Britain's comparative advantage shrank.

What persisted from the British advantage was not the technology lead. It was everything built during the window: the City of London's financial infrastructure, the legal frameworks that governed global trade, the institutional networks, the sterling reserve currency. Those compounded independently of the technology advantage that created the original opening.

AI is a capital amplifier, not a population equaliser

The window does not open for all small countries. It opens for small countries with capital.

According to Microsoft's AI adoption report, the UAE leads global AI adoption among working-age populations at 64%. Singapore is at 61%. Norway is at 46%. These are not random small countries. They are among the wealthiest per capita in the world. They have the infrastructure, the talent access, and the capital to deploy AI at scale.

A small country without capital may not experience the window at all. AI requires compute, connectivity, data infrastructure, and skilled workers to deploy it. Without those, the adoption curve is too slow. By the time capital constraints are solved, the large-population countries may already be at full adoption depth.

If that is right, AI does not narrow the gap between wealthy small nations and poor large ones. It may widen it. The countries already ahead in capital and human development get a productivity multiplier. The countries behind get left further behind.

Output is not productivity

The second cut is less obvious. Producing more does not help unless what you produce is consumed and creates value. A small country that uses AI to multiply its knowledge work output still needs markets that absorb that output.

The UAE solves this through trade and positioning: financial services, logistics, AI infrastructure itself, advisory work, tourism. But those markets are not static. If the countries that currently consume UAE-originated services also adopt AI at depth and begin producing for themselves, the export advantage shrinks.

There is a paradox in here. If everyone uses the same tool to produce more, the output competition shifts but the consumption base does not grow proportionally. Someone ends up with surplus output that has nowhere to go.

The window is real. But it is not a free ride. It requires knowing exactly what you are building while the advantage holds.

The UAE as the live test case

I find the UAE particularly interesting to watch right now because it is not theorizing about this. It is acting.

The UAE appointed the world's first Minister of State for Artificial Intelligence in October 2017, five years before most governments had even started the conversation. Microsoft committed $15.2 billion to UAE AI infrastructure through 2029. G42, Oracle, SoftBank, OpenAI, and others announced the Stargate UAE project: a planned 1 gigawatt AI campus with 200 megawatts coming online by 2026. MGX, the Abu Dhabi government's AI investment vehicle, is targeting $100 billion in assets under management. Early positions in OpenAI, Anthropic, and Databricks are already on the book.

This is not a country building for the AI window. This is a country trying to become the infrastructure through which the window operates for others. Whether that is more durable is actually the more interesting question.

What I am not sure about

I have laid out a framework here. But I want to be honest about where it breaks down for me.

First, I do not know how long the window actually is. The industrial revolution window lasted decades before the catch-up wave arrived. AI diffuses faster. Technology licensing and access are global from day one in a way that 19th-century machinery was not. The window could be much shorter: ten years, not a hundred. If it is ten years, the strategic calculation changes entirely.

Second, I am not sure the British analogy holds cleanly. Britain in 1800 had a manufacturing advantage in physical goods. AI productivity in knowledge work is different. A law firm in Mumbai can use the same AI tools as a law firm in Abu Dhabi. The replication barrier is close to zero. Physical infrastructure and compute are the only real barriers, and those are buildable with capital.

Third, the failure case is something I keep looking for and cannot find. Which small country bet on a technology leapfrog in the last 100 years and lost? What went wrong? I do not know the clean example. Without it, the framework is all upside and no stress test.

The window is open. I am genuinely uncertain about how long it stays open, and what matters most to build while it does.