I think that rather than the future Internet being either centralised or de-centralised, it will be a combination of both. The bits that work best when they are centralised will be centralised, and the bits that work best when they are decentralised will be decentralised. There can even be decentralised versions of things that are predominantly centralised, and centralised versions of things that are predominantly decentralised, if need be. This is the way of the digital world. We can have it all ways!
If we wish to make certain types of information available such that they cannot be censored then a decentralised approach might be appropriate for storing and distributing that particular type of information.
On the other hand there will always be information that is far less critical/controversial to the world, such as cat videos, and these could be centralised, particularly if centralisation offered a resource-efficient way to store and distribute those videos.
Whilst I appreciate that civilisation may end tomorrow I do not think that that sort of thinking has to colour everything we do. Sometimes a nice centralised solution is just the best solution, the path of least resistance. Let’s not make things more complicated than they need to be.
BTW: Personally I want to see less local data processing and more remote data processing, which is a very ‘centralised’ perspective. Why do I want this? Because I want the device that I use to access digital services to be a simple cheap commodity. I want to see our sensor-rich smart-devices replaced with sensor-rich dumb-devices connected to remotely-located cloud computing based data centres.