Computational speed potential

I am developing software for resolving air traffic conflicts in a fast-time simulation. Most resolutions only take a second or two, which is well within the requirements for real time. However, some of the more difficult cases in dense traffic can take on the order of a minute or two, which is not acceptable for real time. (I am running on an HP workstation with 32 cores, and I am using parallel Vector processing)

I don’t need to make my software work in real time, but I would like to be able to claim that there is a clear path to it within, say, five years or so – even if I can’t improve the speed and efficiency of the algorithm. I need to be able to claim that there will be a way to get 90 seconds down to something like 5 seconds or less with improved hardware, and preferably without resorting to something like Spark, which requires a different kind of software.

Certainly a major air traffic control facility will be able to afford a more powerful computer than my workstation, and perhaps that could solve the problem now. Short of some exotic super-computer, are there computers available now that are much faster or have many more cores? And how fast is the core count expected to increase in the next few years? Any ideas?