Linux didn't actually become a better solution until recently (past decade or so). Before that, it was merely a matter of Linux hype & bandwagoning.
Nonsense.
I had a 386 in about 93 or so, when they became affordable to a young researcher. At that point, I had a few choices of installing software: SysV (several thousand $), BSDi (about $1000), or Linux (free). 386BSD was not functional at that time ... it was an experimental system. And getting a distribution from Berkeley's CSRG was impossible for an individual. By the way, please don't think that my criticism of 386BSD is to be taken as a personal dig against Lynn and Bill (whom I know personally, they live close to me, and like have been involved in school volunteering and politics), but as a statement of fact.
Yes, the installer was awful. I started with the SLS distribution (34 floppy disks!), then switched to Yggdrasil, and Slackware. But it worked! And it had Xwindows support (which I needed, I was doing numerical analysis and graphing); on BSD, you needed to have an exact model video card (the Tseng ET4000), since no other cards were supported, and that card was de-facto impossible to buy for individuals, since Tseng had no distribution network.
Here is the real reason Linux became a success: By 94, there was lots of software available for it. For example, I needed a certain data analysis toolkit (CERN's hbook/hplot/PAW), and because Linux was an immediate hit with the CERN and physics folks, it got compiled for Linux within days. I needed a good Fortran compiler, and I ended up making my own contributions to f2c to get it compatible on Linux; at the time, BSD had no Fortran compiler at all (which it didn't need, since the system was intended to be a CS research tool, and computer scientists don't program in Fortran). Also in 94, the first C++ compilers started working reliably, and a commercial GUI builder for Linux came out with a C++ backend. BSD was a niche solution for computer scientists and a few hobbyists, whereas Linux was used in production by then.
By the late 90s, the success of Linux had become inevitable. I remember sitting in the lobby of HP Labs in 99, and Linus was ushered in, he was the guest of honor at a research conference. While a few people in the audience would have recognized Kirk, Eric or Sam, Linus was already a rock star. From a research viewpoint, BSD was ahead (Linux internals were a mess), but Linux already had a giant market share, and more importantly mind share. Just as an example: When the first Itanium chips came out (in about 99, that's when secret prototype chips and boards became available internally at Intel and HP), the first OSes ported to it were ... HP-UX and VMS. Obviously, since those were needed to run on it. At the same time, a very small group (fundamentally one guy) ported Linux to it. While the lab where that port was done was a a heavy BSD user, nobody even considered putting BSD on the chip. What would be the point? At that time, Linux was already used by a massive number of machines, while BSD was a niche product, used for research and embedded products.
That's not "hype and bandwagoneering". The reality is that Linux filled a giant hole in the market, and succeeded spectacularly at it.