Unix was originally given away, but then it got wildly popular and Bell Labs regretted they couldn't capture much of the value it created. When Plan 9 came around, they were much, much stricter about it, and so it was much less accessible.

Unix was possibly the first widely-available operating system with its particular ethos - frustratingly simple and hackable. Plan 9 has much the same ethos, but wound up having to compete with Unix, which had a decades-long head-start.

Possibly the biggest problem Plan 9 has, though, is how well-designed it is. Unix was probably well-designed in its early days, but by the early 90s when Plan 9 came around, it'd been hacked up and contorted to all sorts of use-cases the designers never intended, like networking and graphical terminals. Plan 9 re-unified all these use-cases into a small set of elegant, orthogonal features... but it made anything outside those features even more awkward and difficult. For example, a web-browser doesn't make much sense in a Plan 9 world - you'd just run remote applications and have them display locally, rather than having a special local application that all remote applications must run inside. But it's hard to do anything in the modern world without a web-browser, so Plan 9 isn't a practical alternative for most people.

But Unix was originally built under the Bell System monopoly and they were restricted from selling it, and it wasn't developed as an actual product anyway (Ken was basically disobeying his manager in developing it) so they gave it away free, with source code, to whoever requested it (mostly universities. Early distributions on magtape reportedly included the note "Love, from Ken.")

Plan 9 was post-monopoly, so it was a "PRODUCT" and marketed with various licensing restrictions and you had to pay for it (with maybe a by-seat pricing? Can't remember.) And by then it was competing with all kinds of OFFICIAL Unix variants with real support offered, plus MS-DOS and the first Windows, and also Apple was in the mix.

People didn't get the point of an operating system that made good use of a network that most people didn't use at that time. And compute servers? What is that all about? It just didn't resonate with anybody at the time, although it provided all the abstractions we tend to care about more now.

Some things that stop me from using it more widely are:

The GUI: I just can't relate to the aesthetics nor the heavy dependence on the mouse. Granted that I have used it from the shell, but then I also miss out on some niceties and conveniences.

The security mechanism: I understand the intent of the security mechanisms, but have have difficulty in understanding it sufficiently to configure a cluster of Plan 9 systems using some old HP ePCs.

Device drivers: This is a common problem with most non-mainstream OSes. There are too many incompatible devices out there and thus device drivers are either written for some limited common denominator, e.g. VGA and its variants, or simply for what the developers had on hand. We only need to look at the Linux kernel to see that a substantial part of its development is devoted to supporting the widest variety of devices.

I think that the Plan 9 experiment could be re-positioned as an IoT OS. That is an area that I am very interested in.

BTW: check out Russ Cox's implementation of Plan 9 using the vx32 sandboxing on Linux and OS X (https://swtch.com/9vx).

Most companies can’t survive/thrive the way that RedHat did. Not saying it can’t ever work, but it definitely was a factor in keeping plan9 from getting popular.

There are many reasons for it's failure to gain traction as a general purpose OS, but one major one is that it was never marketed as a general use OS. It was always in the realm of a research OS. So that all but "doomed" it from the start.

Lack of applications, drivers and compelling use cases ultimately caused the project itself to fail. It was also competing against operating systems with huge commercial backing that already had deeply entrenched market share.