How the first number is made
It comes from twenty measurements, five for each area. When a measurement has a goal, its score shows how far Europe has come: the EU wants 20 percent of the world's chip production by 2030, Europe has 9 percent, and 9 out of 20 scores 45. When there is no goal, the score is Europe's share, so a 15 percent share of the EU cloud market scores 15. Each area's number is the average of its five scores, and Europe's number combines the four areas with the weights. The targets behind the index lists all twenty, with the goal, where Europe is today and the source.
A score of 100 means a goal is met, not that Europe is sovereign. The measurements are checked about once a quarter, next in December 2026, and the number does not move in between.
How the second number is made
Every news item we publish from 1 January 2026 gets a score for how far it moves Europe towards running its own technology (autonomy) or towards depending on others (dependence). The score grows with how many of four questions the item changes (who owns it, who runs it, whose law applies, who made the parts), how binding it is, and how big it is. An item we checked against the primary document (Verified) counts in full, and one that rests on reporting (Reported) counts half. The biggest possible item counts 1.0. About six in ten items change nothing and count 0, and most of the rest count between 0.05 and 0.25.
Each month's bar is that month's scores added up, autonomy minus dependence, and the line is the running total since 1 January 2026. Items in the record dated before that day are not counted, and their pages say so. Europe's total combines the four areas with the same weights as the first number, so at the default weights the biggest possible compute item adds 0.30 to it. That is why an area's own total can be larger than Europe's.
What the numbers do not tell you
The first number covers twenty chosen measurements, and each one says what it leaves out under "Evidence and source". The second counts only the news we picked and scored, so a quiet month scores 0 and opposite items cancel. Our picks lean towards European moves, which makes the second number lean positive. The news never changes the first number.
Who checks the scores
A model proposes each item's score. An item that passes the automatic checks goes live with it and carries the badge AI-drafted; an item the checks hold waits for the editor, who releases it unchanged or drops it, except in the archive months from May 2025 back, where a held item was dropped without being read. Archive months the editor has reviewed carry Editor-read, and a correction is written on the item with the original quoted. The method page has the full rubric and the review process.
The data
The numbers and the scores are the two files linked at the foot of this page, for the build you are looking at. Both are open, CC BY 4.0, and the connector and API serve them to your own tools.