Phanerozoic continental subduction zones have produced thick continental crust composed almost entirely of granitoid plutons. While ideas about how plutons form have evolved from models that envisioned large, highly molten magma bodies, the exact processes involved remain debated. Geochronology and seismology have led to the view that plutons form by incremental emplacement; stacked sills represent one type of incremental model whereby granitoids grow top-down by sills underplating their predecessor. Still, many questions remain unanswered, including why sill-like contacts are not often seen in more mature plutons, why the mafic residuum is not observed with many granitoid plutons, why some plutons are compositionally zoned (and others are not), and why geochemical characteristics of intrusions systematically change during magmatic cycles. Here, we propose a hypothesis for the construction of batholiths by amalgamation of plutons formed in a two-stage process. During stage 1, intermediate-composition sills underplate previous sills, forming a moving reaction zone mafic complex that produces a thickening granitoid as the process moves downward. The top of this mafic complex also releases a water-rich, low-temperature silicate liquid (LTSL), which begins ascent by reactive porous flow. During stage 2, the upward flux of LTSL further differentiates the overlying granitoids, increasing silica by 5%-10% and resulting in linear “mixing-like” behavior on Harker diagrams. Multiple plutons building downward in a magmatic cycle thicken the crust, leading to garnet gabbro mafic complexes forming at ~20 km depth. Their high density leads to delamination and net silicification of continental crust. The continuous flux of LTSL up through the arc crust can explain many geochemical spatial-temporal trends found in magmatic cycles, including Pb isotope evolution and increasing Fe3+, and provides a mechanism for addition of water to the upper crust, leading to sustained volcanism through time.