How long does it take to build a Saas application?

Well, a SaaS usually has its main business functionality which depends on the actual product you want to build.
Then there's all the "boilerplate", like user authentication, CRUD API operations, dashboards, notification systems, & more.

And that is where the real difference is. There are a number of boilerplates which sell you all the basic parts, and all you have to do is implement the core functionality.
I reckon it would take you at least 1 week to get the authentication with proper design and error handling done, so that's why I always recommend using boilerplates for big projects.

I have a list of full-stack boilerplates but unfortunately I know only one for Java, it's called Turbovar.

I hope this helps :)

I think it really depends on what the product does and who the audience is. They say that you should have an MPV ready in few months and if you're not ashamed of your MVP than you've waiting too long. Well, I'm not sure I agree with that completely. While I do agree with the notion that you should FAIL fast and ITERATE based on the user feedback, I think there's a good balance you have to strike in creating that prototype/MVP that will satisfy the audience you're targeting.

We are in the B2B market. So we can't just get an MVP out in few months that works to some level but looks ugly. Our audience DOES have some expectations in terms of the UI and UX experience. So in our case, we actually took more than 6 months to develop our MVP. We of course didn't add all the bells and whistles of course but added enough features that would provide VALUE to our end clients on DAY ONE.

Also, coming from a development perspective, always add few months on top of your original estimate. We know that things will break and you'll run into something that you didn't expect, etc. We've all been there. As long as you're keeping your beta customers informed about what's happening, you should be ok. They know that this things happen. We set a realistic expectation in terms of when we would launch and they appreciated that and waited patiently. I have to say, based on the initial feedback we've received, it was well worth the wait :)

It depends on the complexity. "Initial version" may have many meanings.
For example, I was able to develop a multitenant, multidomain SaaS app in under 100 hours, see the journey here: https://www.indiehackers.com/post/im-building-saas-and-documenting-the-process-day-31-35-testing-bug-fixing-deployment-and-finally-mvp-is-alive-6810e1aedf

I have a B2B SaaS I started developing last year. I've run a pilot with a client in the beginning of this year, it worked very well, very good results and now we have a 1-year contract. I did the basic first: React + Firebase + Node in a monolith. I iterated A LOT! I made huge changes. One of the best decisions I've made was to use a NoSQL database because I can change stuff VERY FAST! Another great decision was to use Google Cloud's App Engine so I could deploy my node.js backend code very fast without worrying about handling virtual machines or any infrastructure. Just code as it is. Right now, I don't worry about scaling, coding best practices, CI/CD. I need to ITERATE. And fast. It took me two months to finish the first version of my MVP, another month to make enough changes to be able to run the pilot with my client, two months running the pilot and making some changes. In total 5 months IN MY CASE. And I'm still using the React + Firebase + Node in a monolith and probably will continue to use it for another year until I feel I got enough traction and is time to grow.

It totally depends. I can just recommend starting with the core value of your SaaS.
No user system etc from the beginning. Build it if the core mvp works! But of course keep it in the back of your head that you don't need to refactor a lot to add user/permissions etc

Just use any framework which makes sense to use, you're customers won't care if it's a monkey sitting behind a laptop and doing things manually.

The best way to go, is to validate the idea without code or making landing pages and showing to people. When making https://www.featuremonkey.com/ we validated the idea by making a coda doc which won product hunt no code makers festival. There are other exmaples like buffer using landing page to validate idea.

Now when you code, you shouldn't care much about what framework you use, it's a trap and we often under estimate the time we take to build the product. And we go learn and new framework and then you realise this framework takes too long to do this particular thing which was much easier on the framework you used to know very well. And then end up putting you're hand over you're head and crying.

Just make using whatever that is easiest and then when you have a few customers then you should probably get new problems that the current framework can't handle. Then you change it and move to somehting much better. If you are building a product people really want it's very common you do anyway a couple of rewrites.