Machinepower

The Machinepower Index is an AI sovereignty index that ranks countries on whether they can build, power and control their own AI, and maps the chokepoints and blocs that decide who depends on whom. It scores twelve measurements across three layers: energy and compute, models and talent, and the state's capacity to adopt AI with public consent. A nation's Score is set by its weakest measurements rather than its average.

Machinepower is the strategic capability and capacity to generate and deploy artificial intelligence at scale. It represents a paradigm shift from manpower — the historical foundation of national strength — to sovereign computational and algorithmic capacity. The aim of the Index is to do for that shift what GDP did for industrial output: give governments one measure they can compare across nations.

The Index looks for bottlenecks. AI capability works like a production line: the slowest stage sets the output. Most indices average their dimensions, which lets a good score in one place cancel a bad one somewhere else. This one does not, because a world-class model is no use on a grid that cannot power it. A nation needs all three at once.

The Score comes from the twelve measurements beneath the three layers, not from the layer averages — because averaging hides things. A 10 next to a 3 comes out as a forgettable 6.5, and you lose both facts. Potential is the plain average of the three layers, and the gap between the two is what being lopsided costs.

The exact formula, how much the ranking moves under every other scoring rule we tested, where each number came from and what the Index gets wrong are all on the methodology page.

As of October 2026, the United States ranks first in the Machinepower Index with a Score of 8.30 out of 10, followed by China (8.23), South Korea (7.40), France (6.75) and Germany (6.74). Because the Score is set by a nation's weakest measurements, lopsided nations rank lower than their strengths suggest.

As of October 2026, 54 countries have signed Washington's Joint Statement on AI Opportunity and 37 have signed the founding agreement of China's World AI Cooperation Organization (WAICO). Five have signed both: Georgia, Kazakhstan, Malaysia, Serbia and Vietnam. Of the nations the Index ranks, 30 are in the US-led group and ten are WAICO members, with Kazakhstan, Malaysia and Vietnam in both, and 13 are in neither, among them Canada, France, Saudi Arabia, Spain and Taiwan.

The idea is older than AI. In 1890 Alfred Thayer Mahan, an American naval officer, argued in The Influence of Sea Power upon History that power at sea came from two sources: what a nation could build and keep up for itself, meaning its trade, its shipping and its fleet, and the positions it held on the routes everyone else had to use, the “stations along the road” such as Gibraltar and the Cape. A nation needed both, and neither could be bought with the other. In 1904 Admiral Sir John Fisher, the head of the Royal Navy, named Britain’s positions in a single line: “Five keys lock up the world! Singapore, the Cape, Alexandria, Gibraltar, Dover.”

AI splits the same way. Building and running your own systems is one kind of power, and holding something every other country has to pass through is another, so a nation can have a great deal of one and none of the other. The Score measures the first, and the chokepoint register maps the second.

The Index itself grew out of research for the book, Machinepower: The Realpolitik of Artificial Intelligence, which examines how the ability to deploy AI at scale is becoming the defining strategic capability of the twenty-first century. I am interviewing the people shaping the Intelligence Age — government insiders, think tank heads, executives at the major hyperscalers — to inform its structure.