In 1965, Gordon Moore (one of the founders of Intel) stated that he believed the number of transistors that could be put on the same sized wafer for a minimum cost per transistor would double every year.
In 1975, he changed his estimate to a doubling every two years. For a very long time, that estimate has been incredibly accurate. Because of how semiconductors work, the smaller transistors produced over time inherently used less power while running faster.
Until the early 2000s, that is. That’s when transistors started to become so small and densely packed together that it became difficult for engineers to pack more in and still dissipate enough heat to keep them from burning out.
According to the International Technology Roadmap for Semiconductors, that’s when companies had to start resorting to “Equivalent Scaling,” which is another way of saying “finding other ways to keep performance moving forward when a chipset isn’t moving to a smaller architecture.”