I am not saying we don't need UFB or not moving towards that; merely not the reasons in the minds of many and hence it may not happen the way predicted by those minds. Have you heard of the Moore's Law? Well it is not really a law but the 2nd part of that specifically says while the computing power doubles, the cost of it halves. Obviously my argument does not rest entirely on that simple idea, but it is better to build on top of an existing idea instead of just speaking from the guts, or in some cases personal feelings.
There are computers as small and as cheap as a Raspberry Pi (okay again I am generalising in here, there are heaps of competition of cheap credit card size computing devices against the Pi but I assume people tend to at least know what a Raspberry Pi but not others). Of course, $50 could still be too expensive, and we do have computers as small as a cell right now but to you they could be still too big. So I guess the answer is, we will never have computers that are small enough or cheap enough, which practically makes the question pointless. If size is not an issue, then network computing could be extremely cheap by combining junk together. But of course you need knowledge, not money, to know how to benefit from the parallel processing. If a nuclear missile could be calibrated and launched by combining playstation consoles together, I am sure we could do something less demanding with very cheap hardware. People could learn about all these things without going to fancy university, in fact it is because they don't go to university they could have a chance to learn these.
Anyway I am not against UFB at all. In fact, when there are 50% of our population on UFB let's say on the 30Mb/s plan, then it actually lessen the load of the existing ADSL 2+ infrastructure. For now I get a 20Mb/s download but the ISP has to prioritise the traffic to sort of police usage; but if people move to UFB then there are higher chance I get more bandwidth :)