The sheer number of UTXOs being analyzed at each block. More recent blocks have over 150M UTXOs at this point.
Discussion
I’ve already got all of the data pulled from my node and stored in postgres tables, so core isn’t causing the current bottleneck.