Before rails there was LAMP. That's what Facebook used. Java servlets existed as well. And rented clouds existed before AWS as well. Despite the hype for them on pages like this, neither AWS nor Rails were really big disrupting inventions. The market for both existed before them, and will exist after they've fallen out of favour.
Where innovation happened is in the scaling domain. Terraform, kubernetes, etc, as well as in the SaaS tools you mentioned. But most of those things aren't needed for earliest stage startups. You don't need to be super scalable from day one. You can just run everything from one very powerful box for a while.
Of course this all depends on what your application is. If your product is a cloud based tool to post process movies, it will be a different setting from a CRUD app to plan events.
Having lived through it, it's just loads easier to get something up and going these days. I ended up back in some Java code again recently, and servlets are a) a pain, and b) a general-purpose abstraction. They're adequate for a lot of things, but not particularly great at anything. Whereas these days people have had 15 years to come up with special-purpose code to accelerate all sorts of common activities. My point wasn't that AWS and Rails were the only good things that happened in 15 years. It's that those, which now seem old and boring, were near beginning of a whole wave of innovation aimed at making it easy to have a consumer-grade user experience up and running.
It was that investors aren't going to pay for anything they don't have to. If there's a sufficient stream of web/mobile based startups coming to them who already have some proof, they're going to be much less likely to fund people who don't have that proof yet.
That could change, of course. The pandemic is making it clear how threadbare that ideology is, so perhaps we'll return to valuing things beyond a high rate of shareholder return.
Then I can spin a lot of the other work off to pay-as-you-go SaaS tools. Payments, sales support, CRM, support, marketing, analytics, feature flags, landing pages, A/B testing: all of these are things I don't have to build any more. And now thanks to the rise of things like Slack, Zoom, and Trello, remote and/or part-time work is much more tenable, meaning we don't need an office and full-time commitment to get going.
So yes, VCs are going to take minimum risk for maximum return. But they always have. What's different now is that you don't need to convince some rando that you need $250k in seed money just to find out if your idea might work. If it does, you'll at least be in a better negotiating position. And maybe you won't need to seek investment at all.