Who Climbed and Who Fell in the Global AI Use Rankings

What we measured, in simple words

Microsoft publishes how many people aged 15 to 64 in each economy used a generative AI product, such as a chatbot or an AI assistant, during a period. It calls this the AI user share. We covered the full report in Microsoft's Global AI Diffusion Report, Explained. Here we ask a different question: who moved up or down the league table between the first half of 2025 (H1 2025) and the second quarter of 2026 (Q2 2026)?

A rank is simply a position. If you are 12th, eleven economies use AI more than you. We rank all 147 economies in Microsoft's table, and economies with the same value share a rank. A rank change is the position in H1 2025 minus the position in Q2 2026, so +13 means 13 places higher.

One important filter. Microsoft publishes the same number for several economies in some periods, which looks like a shared regional estimate. We only draw conclusions about the 96 economies whose H1 2025 and Q2 2026 values are their own. The other 51 are still counted when ranking, but we do not call out their moves.

The biggest climbers

There are two different kinds of climber here. South Korea, Japan and Saudi Arabia are already mid-table economies that grew quickly. South Korea added 14.7 points in about a year, the largest gain of any economy with its own numbers. Japan added 8.0. Saudi Arabia added 7.7.

Laos, Mongolia, Thailand and Vietnam are a different story. Laos moved up 13 places with a gain of only 2.4 points, from 6.0% to 8.4%. That is a large percentage jump from a very low start, but a small number of extra people. When a country starts near the bottom, where many economies are bunched, a small gain can shift its rank a lot.

Microsoft's own report also names South Korea as the biggest gainer among major economies for several reporting periods running. Our numbers agree: South Korea's gain of 14.7 points is the largest among economies with their own values.

The biggest fallers (who still grew)

Read this table carefully. Nobody on it lost users. Each economy has a higher number in the Q2 2026 column than in the H1 2025 column. They fell because the countries around them grew faster.

Ukraine shows the effect most clearly. Its value rose only from 9.1% to 9.5%, but because so many economies are packed into the same range, it dropped 26 places. Lesotho moved from 8.8% to 9.9% and lost 18 places. The data does not tell us why these economies grew slowly, and we will not guess.

China fell 6 places (15.4% to 17.5%), but treat that with caution. Microsoft says its current measure leaves out several widely used Chinese AI tools, so China's number is probably too low. It expects measured use to rise most in China when the next report counts more tools.

What happened to the big economies

The table below covers some of the largest economies. A star (*) means one of the two values is shared with other economies, so read that row as a rough guide only.

Three patterns stand out. First, the United States, France, Italy and Brazil edged up by one to two places, which means they grew a little faster than their neighbours. Second, Germany and Canada slipped three places each: both grew by five to six points, but the economies just above them grew by more. Third, India and Australia did not move, and the United Kingdom stayed 8th, so steady growth alone does not change a ranking.

Why a rank can mislead you

• Crowded middle, big swings. 47 economies are between 8% and 12%. In that crowd, small differences in growth can shift a country many places (Ukraine fell 26 on a gain of 0.4 points). Where economies are spread further apart, such as the very top where the UAE is 9 points ahead of second place, ranks move less.
• It is a race, not a score. Your rank can fall if you stand still, and also if you grow fast but the pack grows faster.
• Measurement can change. Microsoft says its next report will count more AI tools. That may lift some economies more than others, so ranks may be revised.
• It measures use, not jobs. A higher rank means more people tried an AI product. It does not say anything about job losses or gains.

If last year's pace simply repeated

This is a simple what-if, not a forecast. We take each leading economy's gain over the last year and add it again. The table shows what the leading economies would look like if nothing about the pace changed.

On this simple extension France and Ireland would be level near 58%, South Korea would move past Norway, and the UAE would reach about 87%. We do not expect that to happen exactly. Adoption usually slows as a market fills up, and the UAE is already at 73.3%. The extension also ignores Microsoft's planned changes to what it counts. Use it to see how fast the current pace is, not to predict next year.

How this connects to jobs

Climbing the table means more people are trying AI tools. That is a different thing from how much of a country's work could be done by AI, which is what our exposure scores describe (these scores are model-estimated, not measured). Across 101 economies, exposure and use are linked at 0.66, but once national income is held fixed that falls to 0.02. We explain that in AI exposure vs actual AI use. For the US at state level, see US states: who uses AI more than their job mix predicts.

What to watch next

The next Microsoft report is the one to watch, because it will count more tools. If China, Russia and other economies where non-Western tools are popular jump, some of the ranks above will change. We will update this page and our data file when it lands. The full data is in our AI diffusion JSON file.

Spotted an error or know a better source? Report it and we will check it against the source data.