The share that did not fall

Eurostat refreshed the quarterly sector accounts on 18 August. The gross profit share of euro-area non-financial corporations now stands at 38.6 per cent, down from a peak of 42.3 per cent in the second quarter of 2021, and below the 39.7 per cent it recorded at the end of 2019. The profit surge that dominated the inflation argument of 2022 and 2023 has been given back in full.

That is not the only European number pointing the wrong way for this book. The Commission's own AMECO series puts the adjusted wage share of the EU27 at 63.3 per cent in 2025 against 64.5 per cent in 1995: a fall of 1.2 percentage points in thirty years. Germany's wage share is at a series high of 66.1 per cent. France is above where it stood in 1995. Compensation of employees was a larger share of EU output last year than in 2015, 2019 or 2022.

The Netherlands is the sharpest national case, and it cuts both ways. CBS reported in July that the labour share of the Dutch market sector was 70.6 per cent in 2025 against 81.4 per cent in 1995, a fall of nearly eleven points. That is a decline on the American scale. But the same release is titled to say that the reward to labour rose faster than company profits last year, and the series has now risen three years running. The Dutch fall is real and it is a story about 1995 to 2022, not a story about what is happening while the models are being trained.

Claim C1 of this book says the split between productivity and pay is real, global and ongoing, and that the labour share keeps trending down. In the United States that is documented: 52.9 per cent in the second quarter of 2026, the lowest since the series began in 1947. In Europe it is not happening. The claim's own stated test for being complicated is a sustained labour-share recovery in economies deep into adoption. Europe has the first half of that and not yet the second, which makes this a warning rather than a refutation.

Two things are worth saying before anyone else says them. The first is that European institutions are different by design. Belgium indexes wages to prices automatically, Germany has codetermination, most of the continent bargains collectively at sector level. If those institutions are doing what they were built to do, a stable wage share is the system working, not a measurement error. The second is why that second half is not yet met: Eurostat records that fewer than one European enterprise in five with ten or more persons employed used any artificial intelligence at all in 2025. A mechanism that has not diffused cannot yet have moved an aggregate.

What survives the correction is the part that was always the argument. The book's standing question is not whether pay is falling behind but who owns the machine, and on that the European evidence is not ambiguous. Sixty per cent of euro-area households own their home and eleven per cent own shares in a company. The richest tenth hold eighty-three per cent of directly held shares; the poorest half hold two per cent. European cloud providers tripled their revenue between 2017 and 2024 and still watched their share of their own home market fall from twenty-nine per cent to fifteen, because the market grew sixfold and the platform belonged to somebody else.

So the honest position is this. In Europe the gains are not yet visibly leaving the wage. They are visibly accruing to an asset that almost nobody here owns. Those are different sentences, and only the second one is currently supported. The watchlist item follows from that: if adoption passes a quarter of firms and the European wage share is still flat in 2028, the accelerationist half of C1 is wrong and should be withdrawn rather than defended.

Which leaves the argument standing on narrower and firmer ground than it had before that release. A European reader can look at the wage share and say, correctly, that the sky has not fallen on their pay. They cannot look at the ownership figures and say the same, because eleven per cent of them hold a share in any company at all, and the platform their employer now rents was bought by someone else while the wage share held perfectly steady. Who owns the machine?