There is no great mystery, its just the lifecycle for each processor
generation. The first product barely works, uses a LOT of juice, needs
cooling, and well its the first board etc. design, so its going to be
large, barely small enough to be acceptable in the market.
Latter in the same generation low power versions become available, and
initial board designs are redone using higher integration (more complex
chips, but fewer of them).
Finally models come out primarily designed for small size, even if the
technology is a generation behind in absolute performance levels. Different
strokes for different folks.