I don't really think any of what you said is actually true.
There are many types of software project where the license doesn't matter at all - what matters is that the software exists. Someone is willing to pay to get it made, and they don't really give a hoot about "licensing". I believe this actually constitutes the majority of software - it's just invisible because no one really has much incentive to release it. But I don't really see how, say, a festival's business model is undermined by making their lineup app open source.
Next, there really is GPL code that's simply too special to ignore. One obvious example is the Linux operating system. Designing an embedded device? You're almost certainly using Linux. Maybe Linux doesn't do quite what you want - say, support for a given bit of hardware. Guess what, you can't ship your patch without publishing it - and now every Linux user gets to use that hardware, if someone's motivated enough to merge it into mainline. This is a real thing that happens all the time.
We're not there yet (and we may never get there if people stop believing in the GPL), but you can imagine that if we get a world where all the important software is GPL, the sheer mass of it will overwhelm any other licensing considerations. "You can either participate in our ecosystem, or you can rebuild modern computing from scratch. Your call."
If somebody doesn't care about licensing, then it's trivial to convince them to use open source whether or not they want to tightly integrate with GPL code, so the GPL aspect isn't increasing the amount of open source software.
Your hardware driver example is pretty much the only example I can find where somebody is actually incentivized to release code, and willing to go against other concerns; so I appreciate that example! However it's certainly not universal for all hardware, there's plenty of binary blob drivers out there on Linux unfortunately. GPL doesn't need to universally make all software open source for it to be more successful, so I think it's important to look for the concrete overall results.
Let's contrast GCC and Clang here. For years, advancements in open source tooling was hindered by GCC's refusal to make improvements that could potentially be used by closed source software. Clang comes along, and all of a sudden an entire open source, tinkerable compiler/tooling/editing environment opens up, precisely because both open source and not GPL. People are willing to pay to improve it because they have the freedom to do what they want with their improvements, including the freedom to hold them back.
Despite not having the stick of the GPL, MIT-style license resulted in a better ability to tinker with the code, expand it to more areas, and improve software.
I don't see GPL as a crucial aspect for getting to that world where we have access to the important bits.
My goal is to enable as much as can be done with software that also has source available. There's a different optimization goal, that says "let's make sure that all software out there has the source available." This is a drastically different goal, however, because it can be achieved by creating less software capability overall, and just eliminating all the growth in capability that originates first in proprietary software but then gets reimplemented as code-available software. IMHO, the GPL seems to be aiming for the "any software out there also has source available" goal.
20 years before clang, MIT style licenses resulted in a huge number of duplicate, incompatible systems along with a general inability to improve software, to the extent that Windows became the de-facto industry standard OS.
There's a reason gcc became the best of breed compiler before llvm came along.
There are many types of software project where the license doesn't matter at all - what matters is that the software exists. Someone is willing to pay to get it made, and they don't really give a hoot about "licensing". I believe this actually constitutes the majority of software - it's just invisible because no one really has much incentive to release it. But I don't really see how, say, a festival's business model is undermined by making their lineup app open source.
Next, there really is GPL code that's simply too special to ignore. One obvious example is the Linux operating system. Designing an embedded device? You're almost certainly using Linux. Maybe Linux doesn't do quite what you want - say, support for a given bit of hardware. Guess what, you can't ship your patch without publishing it - and now every Linux user gets to use that hardware, if someone's motivated enough to merge it into mainline. This is a real thing that happens all the time.
We're not there yet (and we may never get there if people stop believing in the GPL), but you can imagine that if we get a world where all the important software is GPL, the sheer mass of it will overwhelm any other licensing considerations. "You can either participate in our ecosystem, or you can rebuild modern computing from scratch. Your call."