In other words, it is always easy to think it is too late when looking back. But it is never too late if we try to look forward and imagine the things that have not invented yet and which will be around in 10-20 years time.
Marc Andreesen said glibly that Windows was nothing more than a poorly debugged set of device drivers. The technical press parroted it without any follow-through. When Microsoft came out with Internet Explorer, new versions of Excel, Word, and IIS, I knew they would win. Because device drivers are hard, nobody wrote them better, and Windows was good enough for millions of customers.
I think Microsofties low-key regarded me as a bit dull, because they believed the media had no agenda. I simply looked at the (public) accomplishments of my much younger coworkers and trusted my own observations.
To me, it's been "obvious" for the last US$2 trillion of market capitalisation growth that Apple's going to face monopoly investigations; it's "obvious" that Facebook is, in a deep fundamental sense, not even capable of being compatible with GDPR; and it was "obvious" to me in 2009 that the first true self-driving car sold to the general public (by the quality standard "doesn't come with a steering wheel and that's fine") would be around 2019 or so.
There's been a few places I've worked with young energetic co-workers (and I have been a young and energetic coder myself), and… in many cases those groups failed to make enough money and disbanded.
One result of a lifetime of not being able to predict the markets is the realisation, a decade ago, that my preferences in many areas are almost anti-correlated with the modal buyer: I like what normal people dislike or even loathe, and dislike or even hate what they prefer.
Predictions of how it will be are always worth writing down, to see how correct they are.
Microsoft and Apple are huge today, but it's entirely conceivable that both are shells of themselves in a decade's time.