For all the constant buzz around AI, broad-based announcements about how AI has boosted profits are noticeably missing. The reason for this loud silence: it hasn’t happened—yet.

AI is not an immaterial vapor hovering above the economy. The models that will power future change run on a foundation of physical inputs: cables, turbines, chips, and copper. These are bought and sold all over the world. This is why the one area AI is having a measurable effect on trade.

According to recent research from the McKinsey Global Institute, shipments of the chips, servers, and networking equipment needed to build data centers increased by almost 40% in 2025, accounting for a third of global trade growth. In the United States, which accounted for approximately half of new data-center capacity in 2025, AI-related trade rose by two-thirds, or an estimated $220 billion.

The reason AI’s impact on trade is so wide-reaching is that while a data center might be located in Virginia and use American-designed semiconductors and large language models, it is ultimately a citizen of the world, home to chips fabricated in Taiwan and extreme-ultraviolet lithography machines made in the Netherlands. These are paired with Korean memory chips, which may be packaged or tested in Southeast Asia. Mexican turbines and Chilean copper power it up.  In other words, that data center is the end product of the complex workings of global supply chains.

The simple fact is that even if AI is the technology of the future—and it is—it will still depend on history’s traditional engine of prosperity: trade. For the United States, that fact leads to several conclusions.

Domestic capabilities are the foundation

American ingenuity was a genesis for AI, from the invention of the transistor to the development of the transformer, which enables large language models. To this day, U.S.-based firms account for most global semiconductor sales and have leading positions in research, design, and process technology. These capabilities make the United States central to global AI supply chains, not just at the receiving end. The point of strengthening domestic capabilities, then, is not just to make more AI inputs at home. It is to remain an indispensable player throughout.  

Through the 2022 CHIPS Act and other initiatives, the U.S. is making concerted efforts to increase production all along the semiconductor supply chain. Having a home-based supply of chips makes the AI economy less vulnerable to disruption. It also integrates American firms into global innovation. Even when components are made in the USA, however, that does not mean U.S. companies are going it alone. Industry leaders like Taiwan’s TSMC and South Korea’s Samsung, for example, are investing tens of billions in U.S.-based production. 

The U.S. needs trusted partners

This is why the U.S. still needs trading partners it can trust. 

Given the rising demand for AI-related goods and services, it is essential to ensure supplies of essential inputs. This cannot be taken for granted. For example, high-performance memory chips are almost exclusively produced in South Korea and ran short in 2025. Furthermore, AI-related trade restrictions are spreading. Washington has tightened controls on China’s advanced semiconductors, high-bandwidth memory, and chipmaking tools. For its part, Beijing has restricted exports of rare earths and related items.

For steady, reliable access, the United States needs trusted suppliers. That is leading to a new geometry of globalization. MGI has created a metric, which we call “geopolitical distance,” that defines the extent to which trading partners are aligned on important global issues. Since 2017, geopolitical distance has declined markedly, with China, the European Union, and the United States all reorienting toward closer trading partners.

The importance of trade

The truth is that data centers need chips and servers—but they also need, literally, nuts and bolts. Plus, air conditioning, generators, cement, lightbulbs, and even coffee cups for workers. So while Japan, South Korea, and Taiwan are essential at the top end, strong relationships with markets like ASEAN and Mexico, which are strong in the middle, also matter. 

It’s not just about goods trade, either. Ultimately, we believe AI leadership will also result in more trade in services. For example, U.S. multinationals may train models that are then hosted in data centers built by American tech giants, serving clients anywhere.

More broadly, to stay a global leader, the United States needs all kinds of connections, not just those related to trade. Capital is one, and indeed, foreign capital already funds important U.S. assets, such as semiconductor labs and data centers. Intellectual property is another. It is important to ensure that knowledge can be shared, whether through licensing or through people. 

How big will AI be? It’s impossible to say. What can be said is that AI is essential to productivity growth and for the United States to maintain its global economic leadership. 

The bottom line: to build a prosperous, AI-driven future, the United States needs a little help from its friends—and good reasons for those friends to keep choosing the United States.