Major network transitions tend to be hybrids. So it will be with 5G. Though most believe 4G will continue to anchor “coverage” requirements, 5G clearly will represent upside for capacity requirements and low-latency apps.
In other words, 5G will largely complement 4G, not displace it, in the early going.
In substantial part, small cell deployments will increase for two reasons. Small cells are virtually required for use of millimeter wave spectrum, as the signals simply will not propagate as far as sub-1-GHz signals.
On the other hand, millimeter wave signals are prodigious bandwidth tools, using space division to reuse frequencies, as well as the inherent properties of millimeter wave signals.
All other things being equal, millimeter-wave signals have an order of magnitude more bandwidth potential than sub-1-GHz signals have (potential symbol transmission in any given time frame).
Between 2019 and 2023, some early 5G adopters are going to deploy 5G-NR small cell systems that use 4G network signaling, but the 5G air interface, allowing an early launch of 5G before full 5G networks can be built.