The number of second level domains may not be the most accurate way to measure the value, the performance or the quality of a new gTLD domain. Is it better to have many domains, but all of them are parked and will most probably not be renewed, or rather to have a few domains that are really used ?
The question is hard to answer, but it is also hard to assess the type of use of TLDs.
If we look at the total number of domains, the ranking of the new gTLD is fairly stable. Even though, since august 2014, more than 500 new TLD have come to live, we still see the same top 20 ranking more or less.