The article also doesn't mention an obvious bias, that is, if a startup starts in Europe, there are a lot of incentives for that startup to move to the U.S. early in its life cycle.
In that scenario, the startups who become successful (and who originated elsewhere) all get counted as U.S. successes. This would vastly undercount both the total number of startups elsewhere as well as the number of successful startups.