Middle Powers Won’t Pick a Side. They’ll Pick a System.

Mark Kennedy

November 17, 2026

Adapted from remarks at the Second Krasno Global Policy Forum, UNC-Chapel Hill, November 17, 2026

In January at Davos, Canadian Prime Minister Mark Carney warned that middle powers must act together, because if they are not at the table, they are on the menu. At the Krasno Global Policy Forum at UNC-Chapel Hill, I offered a different frame. In a digital world, middle powers are not on the menu. They are the customers. And their choices will decide the most consequential competition of our time.

The end of ambiguity

For thirty years, a manufacturer in a middle-income economy could build one product, in one factory, with one set of suppliers, and sell it to the whole world. That era is closing, because supply chains are becoming visible.

Digital tracking, provenance requirements, forced-labor laws, carbon border adjustments, rules of origin and export controls on advanced chips mean that every component increasingly carries its own passport. Goods routed through a third country to disguise their origin are easier to spot, and both Washington and Brussels are looking.

Visibility forces choices. Middle powers will increasingly run two supply chains side by side: one oriented to China, and one oriented to the United States and its partners. That means two supplier lists, two compliance regimes, and sometimes two production lines. It is expensive. But it is also leverage, because both systems need them.

Goods can be split. Digital stacks can’t.

In physical goods, a middle power can serve both sides. In the digital stack, it cannot do so easily.

When a country chooses whose cloud it runs on, whose AI models its businesses and ministries use, whose payment rails carry its commerce and whose equipment forms its telecom backbone, that choice is sticky. It shapes where the country’s data lives, which tools its developers learn, and who can see into its networks. Those decisions last a decade or more. That is why authoritarian and democratic systems alike will court them aggressively, with financing, bundled hardware, and competitive pricing.

The real AI race is a race for adoption

We often talk about the AI race as if it were only about who builds the smartest frontier model. That matters, but the decisive contest is adoption.

AI runs on a flywheel. More users generate more data. More data builds better models. Better models attract more developers and more applications. More applications attract more users. Whoever spins that flywheel fastest pulls ahead, and the advantage compounds.

The United States and China cannot settle that race on their own. Most of the world’s next billion internet users, and much of its next generation of developers, live in India, Indonesia, Brazil, the Gulf, Africa and Southeast Asia. Middle powers are the swing vote in the AI flywheel. Win them, and your system becomes the global default. Lose them, and even the best model in the world becomes a niche product.

What wins their choice

If middle powers will choose a system, democracies need to understand what wins that choice. I see four things, and we need all four.

Capability comes first. Frontier leadership is not just a matter of prestige; it drives adoption, and cybersecurity shows why. The most capable AI models will be the first to find software vulnerabilities, and the first to defend against them. A country that partners with the frontier leader gets early warning of cyber threats. A country that partners with the laggard learns about them after the attack. For a middle power protecting its power grid, its ports and its banks, that difference is decisive. The frontier race and the adoption race are not separate contests. One feeds the other.

Next, trust they can afford. The democratic edge is AI that is safe, kept under human control, and respectful of people’s data. But China is offering cheap open models and financed infrastructure. Trust that a middle power cannot afford is not a strategy. Democracies must make the trusted option the affordable option.

Then, partnership rather than dependence. Middle powers do not want to be treated as markets. They want to be co-builders, with local data centers, local talent, a share of the value and shared cyber defense. Co-financing with allies in Japan, Europe and the Gulf, alongside American development finance, can make that real at scale.

And finally, reliability. Every time the West is unpredictable, middle powers hedge, and hedging is the other system’s opening. In this contest, reliability is not a soft virtue. It is a competitive asset.

The systems advantage

The competition of this era is not only between nations. It is between systems: of technology, capital, production and alliances. The side whose system is most capable, most trusted and most widely adopted will set the rules for decades.

Middle powers understand this better than anyone. They will not pick a side. They will pick a system: the one that is most capable, most trusted, most affordable and most reliable. The task for the United States and its democratic partners is to make sure that system is ours.

Author

Mark Kennedy

Director - Development Research Institute