The rise of AI, is it a bubble? Nobody can be sure. But one way to answer that question is to ask a more approachable (and more interesting) one: What kind of global economy would have to emerge for today’s stock market prices to make sense?
Let’s think about the core of companies that are leading the rise of AI: Nvidia, Alphabet, Apple, Microsoft, Meta, Broadcom, Tesla, OpenAI, Anthropic, SpaceX-xAI and Amazon Web Services. Together, they represent an extraordinary market bet. Under a conservative scenario – in which, by 2036, these companies are trading at a price-earnings ratio of 20, earning margins of 20% and receiving 65% of their incremental revenues from abroad – in a decade they would generate approximately $2.4 trillion in additional annual revenues from abroad. Those revenues are roughly equivalent to total U.S. goods exports today and are more than double the U.S. current account deficit.
It is true that the United States will have to import certain inputs (for example, semiconductors) to provide these AI services. And not all the profits will go to Americans, since foreigners also have shares in American technology companies. But this second effect is smaller: foreigners, according to existing datathey only own between 15% and 20% of the S&P 500 stock index.
These shocking figures should lead us to rethink the current global macroeconomic debate. For years, the discussion about global imbalances has revolved around a central concern: how long can the United States continue to run large external deficits? But if the markets are more or less right about AI, the more pressing question is how the rest of the world will reward American capital for AI services.
This is a radical change. The rest of the world will not only be asked to recognize America’s technological leadership. You will be required to pay for it – year after year and on a large scale. Let’s forget about the industrial export boom that President Donald Trump keeps promising. The big money transfers will go to a relatively small group of companies that control the big language models, chips, cloud infrastructure, software ecosystems and complementary platforms on which the AI era depends.
How could the rest of the world pay?
Countries finance their imports with exports of goods and services, investment income, asset sales or loans. But today the United States is moving toward greater protectionism, making it difficult for foreign economies to obtain the dollars needed to purchase American goods and services. That represents an obvious contradiction. The United States cannot demand that the world transfer huge sums to its leading AI companies and at the same time restrict the channels through which the rest of the planet acquires purchasing power.
This tension has far-reaching political implications. Silicon Valley and Trump’s MAGA movement may be part of the same domestic political coalition, but their interests are fundamentally misaligned. If the stock price of leading American AI companies depends largely on the future growth of their foreign revenues, then for them access to foreign markets is not secondary, but fundamental. And foreigners – who are expected to buy American AI products and services on an unprecedented scale – must also be allowed to sell something in return.
Now, this story does not end there: a massive expansion of the supply of US exports is only viable if the relative price of those exports falls – that is, if the terms of trade of the rest of the world improve with respect to the United States. That makes the political dilemma facing the MAGA coalition even less attractive.
From this perspective, much of today’s trade debate already seems outdated. The obsession with bilateral deficits and tariff policies seems curiously parochial – and increasingly irrelevant – compared to the prospect of trillions of dollars in recurring payments to a handful of American technology companies. If current prices are broadly correct, the next phase of globalization will not focus on the US trade deficit, but on the need for the rest of the world to pay for their access to US-owned AI infrastructure.
Then there is the tax issue.
The rents implicit in these stock prices are enormous. And they benefit companies that, together, today employ less than a million people and, therefore, have a market value of 23 million dollars per worker. This is not an example of widespread job creation. It is an example of how a small group of people can make up a good part of the future income of the rest of humanity.
That makes an increase in taxation not only possible, but politically indispensable. The US government will want a piece of those juicy new revenues. State and local governments that host data centers will want a piece. And foreign governments will want a piece. The current disputes over taxes on digital services will come to seem like a minor prelude to a much larger fight over who can tax AI income.
And taxation directly influences stock prices. If these incomes continue to be taxed lightly, then current prices could be justified. But if governments act as they typically do when faced with high, visible, and politically vulnerable rents, then one of two things will happen. Either these companies will have to generate even more revenue and pre-tax profits, or shareholders will ultimately receive less than current prices imply. In the latter case, it is more difficult to rule out the hypothesis that it is a bubble.
The geopolitical implications are even broader. The rest of the world is unlikely to passively accept a situation that involves very large recurring payments to a small number of US companies. Countries will try to reduce their dependence on the United States. They will subsidize national producers, impose local content requirements, favor national leading companies in public procurement, tighten competition policy and design new taxes and regulations aimed at taking part of the profits obtained. If current stock prices are correct, we can expect an era of American corporate dominance and, at the same time, growing resistance to it.
Far from predicting the end of American power, this scenario points to a new form of such power. In the 20th century, American power was based largely on its scale of production, its military reach, and the strength of the dollar. In the 21st century, it could increasingly rely on ownership of indispensable AI infrastructure. The challenge for the world will be how to finance access to it. The challenge for the United States will be to sell the services of that infrastructure around the world, facing nationalist trade policies and the inevitable pressure to tax extraordinary income.
Maybe current prices are a bubble. Maybe markets are right about technology, but wrong about policy and taxation. Or perhaps they are correctly anticipating an economic order in which a handful of American companies command an unprecedented share of global income. But if this latter scenario is what the future holds, then our current public debate is focused on the wrong questions. The defining issues of the next decade will not be tariffs, deficits, or American decline. They will be how the world rewards American capital that produces AI, how long protectionism can survive that reality, and who will tax income.
*This article was originally published in Project Syndicate.
