SNA was cloud computing 30 years ago and it failed because centralized computing cannot provide the creativity needed. The if you look at an iPhone 6 and compare it to a 1970 IBM 360, you will see the progress is moving functionality into silicon chips. While there is a group of people who think central process is future, the reality it will be the ability to create custom silicon. When you understand the functionality of running a calendar system off of an iPhone, instead of on a main frame, then you understand the cloud can really only be a communication network, not central processing network.
If you look at Adobe's model, you have conform to the least common denominator as their cloud software tries to accommodate the masses. As soon as someone offers custom silicon for individual needs, they will fail like SNA did. In 10 years, if I want to render 5K video to three formats, I will be able to order a rendering cube. Adobe will be trying to get everyone to conform to there least common dominator software.
The cloud may become backup storage, but it will not become a processing engine. The latency of the cloud will render that useless. Yes, you could run sna applications on the cloud quite effectively, but you will not be able to run future application on the cloud efficiently.
Lets say I want to render a new paint job for my video paint on my car. At 5k resolution, you will have to have silicon in the car to render the new colors and a high speed connection from the workstation to the rendering engine in the car.
If you think of trying to do that over a cloud, it will never work.