Well, taking a look at Wetiken's sig:
That still is 3 and a half years that we have built and experience and also shipping. I'm not going to tell you what the next innovations are, but our roadmap is full, because to continue to improve transistors, you have to make substantial improvements. And we plan to do that, while other people are working on perfecting their FinFET devices, and we're gonna be moving on to looking at what comes next. --William Holt, Intel
One can hold out for a break-through that some post -FinFET device might offer better frequency scaling. It seems like every 'magic' material that comes along with so much promise, like vertically stacked graphene layers, which could provided huge boosts in electron mobility, come with their own set manufacturability problems that might just rule them out for many years to come.
It looks like thermals and QM effects are likely to make it even harder to maintain, never mind increase, clock speeds. Intel has been extremely creative so far - but I expect for cost reasons, they won't push their tech any faster than they need to, which is to say - 5-10% per generation.
The new client server model, cloud computing, etc., is driven my mobile clients and servers (handling both storage and remote compute duties) and wireless bandwidth to create a new distributed computing model compared to the standard Server->LAN->PC basic tiered system. This is what is increasing driving CPU/SoC requirements - and it changes the ball game.
The next big break through in total throughput may just have to wait for quantum computers - but that's a very different story which will transform the computing ecosystem (hardware, interfaces and, at the very least, hardware abstraction layers because the software ecosystem will likely lag the hardware developments).