Things to consider will you need a lot of custom database management? Are you comfortable deploying MySQL or another DB from command line? Are there tools in AWS that you will benefit from using - will they save you time compared to building them yourself, e.g. SQS versus managing a queue system.

If you are prepared to manage these are yourself, feel comfortable doing so, and your tech stack won't benefit from the AWS tools too much, using a VPS might be the way to go.

That said you're going for MVP - and you have 6 months to do it. Get going, move fast and break things!

I've done a fair amount of "command line" ops so I'm comfortable with it but the question what will migration be like when we do move to an iaas?

It depends what your infrastructure consists of. If it's a db with infrequent writes it's going to be a lot easier to move than a db with a lot of writes.

You could also host on VPS and use parts of AWS e.g. S3 if you require asset storage, like photos, videos or other files.

Do you have much insight into what your infrastructure will look like at the start?

Are you suggesting that you have done zero development yet? Develop locally until you need things to go "live". Don't buy AWS on day one, you won't utilize it. Save the money. There are many great ways to develop locally, even with teams. You can go far by utilizing dependency management services, like npm, etc. You can go far by utilizing local database init scripts, etc, for consistent dev environments after a fresh repo clone.

maybe mvp isn't there right word. we have demos we've developed locally and we're going to start pilots.

You should just start with a managed setup; a VPS could work but AWS gives you an overhead to manage and be aware of.

While your building up; you may switch to AWS once you need more scaling ability. But if you pick some hosting provider like Internap you can get hardware pretty fast and will be cheaper in the longrun than any public cloud.

Sad; classic hosting is now trying beat the public cloud. The public cloud is trying to get everyone on it and kill classic hosting.

It will never die however; because even Amazon needs a datacenter.