Not going far enough ahead. Today's computers, for anything other than gaming, video editing and mass productivity work (like protein folding simulations), are so fast that for any standard task they are well, well in excess of its requirements by order of tens of thousands. In 20 years, or 30, do you really think most personal computers as we know them, even for professionals, are going to resemble the same old boxes and motherboard designs we know now? How do you know all that power we now value won't be in those days taken completely for granted?
What sorts of tasks do you regularly do that you even need a high-end system for, other than gaming (which has already well reached the point of diminishing returns with respect to image quality vs. cost and will only get worse as time goes on)? Even if there are some you do with reasonable regularity, what's stopping you from, say, renting a day on a server and getting those computers to do it for you?
You really don't want to use me as an example here, bro. I run out of memory on a 16 GB RAM desktop on a regular basis, by doing nothing but loading a bunch of data into memory to run tests that I can't be bothered to rewrite to run out-of-core.
That aside, you have a very naive view of how software development follows advances in hardware. People always think "There's no way anyone would need more than X Megabytes of RAM.", and are consistently proven wrong. There's a quote, I think it's from Silberschatz and Gavin, not absolutely sure, some book on operating system design for college students in any case, which goes roughly like "An increase in available system memory has no observable effects except for the increase in average application memory usage." - and that's how it's always been. You'd think 4GBs of RAM would be enough for someone working mostly with Office, a web browser, and some specialized qualitative data analysis software, right? Well, my colleagues at work consistently manage to choke their laptops by having a zillion documents, spreadsheets, outlook windows and browser tabs open.
An increase in hardware capability is followed by software adding more features, improving user experience (browser is more responsive because it now keeps everything in memory instead of caching some of it to disk), improving time-to-market by using tools which result in less-optimized code that is easier and faster to write (as we can see with the rise in popularity of dynamic, high-level languages like Python or Ruby). You can argue that this will eventually stop, but that is, again, a naive view. It's like a XIXth century politician thinking how everyone will be happy once they gain the luxury of hot running water. Expectations and needs grow as they are fulfilled.
I can believe that desktop computers will someday become obsolete, but at that point, we may as well all have chips implanted in our bodies, running off naturally generated heat. I certainly wouldn't expect it within the next ~50 years. Especially with VR on the horizon, that tech needs a ton of local processing power to work, so if anything, we'll see more boxes rather than less in the near future.
