My explanation is that the computers just got easy enough
to use to move down to the poorer social groups for whom
a "sweet price spot" is lower, and also that there is no
software that would make a $2500 machine universally
more attractive than a $1000 machine, to justify the additional
expense for all people - while alternative ways of spending
the difference promise greater utility.
This process will probably continue, so we may soon see
easy-to-use general-purpose PCs for $100 that kids can carry
in their backpacks, etc.
There may be a new life in higher-level part of the market after
the next waves of software products [?] hit the market.
Another interesting question is, why the computers are not
getting more expensive? Even the richest people wouldn't
buy a $50,000 machine, let alone a million-dollar machine.
This machine would just take more space and require more
maintenance, while giving no perceptible advantage
If we see applications that can work well on a mass-market PC, but would perform much better on more and more expensive machines, people would buy these machines. (Though I am not sure even here; unless this power is needed locally all the time, it may make sense to dispatch computational requests to the specialized distributed machines rather than execute them locally). What would these applications do? Speech and image recognition? AI? All-purpose personal extensions?