I'd suggest Rails. You'll be able build a web app or an API. You'll learn about SQL. (Rails is a framework for web apps - which are generally useful to know about if you're not sure what you might do later). For deployment, I have had a great experience with Heroku, which is probably a bit easier than AWS. I'd suggest ignoring front-end frameworks like React for now - you can pick them up later, if your use case requires it. And you can probably get a decent MVP done without them. At the beginning you want to optimize for speed and simplicity, so that you can iterate quickly.

I agree w/others here that picking a good problem to solve is the MOST important (along with having a decent idea of how you might find customers) - and implementation/selection of technology less important. That said, when I was thinking of starting a company, I learned Rails first (I used DHH's Rails textbook)! For me, it was useful to learn web programming so that I could understand what web apps were even capable of doing. It's really helpful when you're picking a problem to work on to at least have a very vague general idea of how software might solve it! So for me at least, it helped inform & inspire our brainstorming/ideation discussions. And since we did end up using Rails to build our MVP, when my cofounders and I all quit our jobs to work full-time, we could start building right away.