Moore's law states that no of transistors on a given area doubles every 18 months. Electronic components are typically made by photolithography. So, Moores law basically limits the resolution of photolithography. Why resolution of photolithography evolved gradually and not abruptly?
There's a adage: you have to crawl before you can walk.
You may as well ask why supersonic aircraft weren't developed first, before people started experimenting with gliders, the Wright brothers developing the first powered aircraft, followed by propeller powered biplanes, then monoplanes, then jet aircraft and finally supersonic jet aircraft. All this over the course of some five to six decades.
Every technology has its limits, then someone develops a better way and improvements are made and the technology advances. The same applies to photolithographic techniques that are used to make electronic chips. The technology used to develop the first chips couldn't make the chips of today & the engineers of the day didn't know how to develop techniques to make the chips that are made now.
The component density on chips made by photo lithography is directly dependant on the wavelength of the light source used and as shorter wavelengths equate to higher photon energies you need to develop practical sources for ever increasing frequencies and the masks and dyes to go with them,
Equally you need more and more precise and clean manufacturing plant. Once you get to UV lithography a stray hair or skin cell is like hitting your machines with a big hammer.
Part of this is the investment required to manufacture and develop the tooling required for example you need to be able to create very hard vacuums and incredible precise mirrors.
Moore's law is not a scientific law, it is only a business model (a roadmap).
It has become very famous because of its accuracy especially during the time when microelectronics was making its first steps.
When you are a business man and you manage to predict accurately how technology will develop you will have a much better chance to also estimate how markets will develop.
Also since ~2009 the resolution of photolithography is not linked Moore's law anymore.
Here is a simplified chronology:
- When microelectronics was making its first steps the most successful business men investing in microelectronics were those who followed this estimation from Moore on how microelectronics would develop.
- Following the success this estimation became widely known and got a name: the Moore's law. This law then started to also drive the investments and would serve as a kind of expectation/agreement between the companies and investors.
- there was several attempt to push the technology faster, especially as the microelectronic markets were continuously growing up. If someone could bring a technology a little faster there was a possibility to make a lot of money. All of these attempts somehow failed or nearly did. There was a small acceleration with the nodes 90nm->45nm but the technology limits then catched up, together with the Financial crisis of 2007–2008.
- Since the technology node 45nm (or 32nm depending on the opinions) the density of devices per area is not anymore linked with the patterning resolution but with the patterning technique. You can add to this a small set of submicron effects in the devices taking more importance: It is now more the investments that drive the technology than the other way around.
Nowaday, a new technology has become very expensive to bring on the market, and is therefore only possible for the richest companies and the smaller investors are looking toward a some end point for this advance. But no worry, In the meanwhile they created the 'More-than-Moore' to drive new investments.