In my experience, you learn useful things faster when running your own company, compared to working for someone else. Pretty much by definition, you tend to specialize when working for someone else. You can get unusually good at a certain thing, but you miss out on other things that you could be learning. The thing you specialize in gets boringly easy (from the perspective of what is good enough for the employer), and everything else remains a black box.
As a software engineer, the challenge is learning sales, and the art of distribution in general. If you don't want to do sales, avoid B2B software startups entirely, because that is pretty much always the bottleneck unless you are doing some kind of fundamental innovation.
I don't know anyone who was so good at technical things that a market formed around them and their technical skills, where I know several people who basically knew they wanted to start a software company, started it, slogged in obscurity for 6-12 months thinking product was the most important thing, and then either gave up and got a job again or learned sales in order to survive, and then after another year or so had a business with more fulfilling work and income than the job they left.
The key, regardless of how technical they were, was figure out how to be passable at sales before they ran out of money. Passable sales and passable product ability in practice seems to run circles around great product ability but insufficient sales ability (anecdote, at least for B2B since no one I know has a successful B2C startup). So that is probably the biggest observed blind spot for a software engineer.
The other thing is that a particular idea might have a shelf life, but the concept of starting a business doesn't. Besides having kids or impulsive lifestyle inflation, you can pretty much put off or pull forward starting a business with impunity and you just accept the trade offs. Having a job is without question an easier way to make money though, it's just a harder way to make lots of money or have deeply fulfilling work. I suspect much of the problem actually isn't the job itself, it's the nature of specializing.