That's a good point, and yes I agree that could have explained the statement. In fact the issue was a similar one of using a different reference population to define the percentiles.
Based on my memory of some follow up discussions and a little googling this morning, the CDC uses a criteria called BMI-for-age for childhood obesity. Within this metric "obesity is defined as a BMI at or above the 95th percentile for children and teens of the same age and sex." It certainly be rather hard to end childhood obesity if the top 5% of children are always going to be defined as obese.
In fact, I later learned that all the percentiles they use are defined relative to a reference population of "US children who participated in national surveys from 1963-65 to 1988-94". Which is a reasonable sounding approach. The problem is no one talks about the reference population, hence the statements that, at first glance, appear mathematically self contradictory.