I would focus on understanding future customers' expectations and have a simple prototype made in tools such as Sketch before I would spend any time on the technical foundation.
Peakon started, for example, as a PDF that the founding team brought to potential customers and investors before making any actual development.
You can make a lot of false assumptions in the beginning in regards to what the market is demanding, so it is crucial to be able to test it quickly.
If I could choose, and I guess you can in this thought experiment "The Perfect Startup", then I would always go for a setup with a few, skilled developers and an agile approach where you will be limited in time and resources and therefore need to expand the platform with features in order of priority. Then you'll have control over the quality of what is being developed, which is a huge advantage in the long run.
I know many startups/scale-ups (none mentioned, none forgotten) that are reaching a phase where an incredible amount of resources are needed to rebuild or even throw away large portions of their platform because they have under-invested in quality from the start. And there is no time in a company's life where it is practical to have to stop and pause the development!!
If you know that you plan to make a mobile app later, you have to think about the architecture from the start, so you get a relatively clean API with functionalities that can be used for both web- and mobile frontend, and not, for example, get a backend service that returns HTML to the front end.
Scaling is about understanding where the pressure is expected to come from - is it in many concurrent users? Peak loads? Having to handle large amounts of data – or what scale is actually needed?
You can get far by choosing some sensible 3rd party services from the PaaS provider you use (Amazon, Google, and others) who can help with the scaling – it just requires an understanding of where the pressure will be.
Choose one of the major platforms (Amazon, Google, Microsoft) and use that platform's services for everything you don't need to be an expert in. There is no need to reinvent the wheel when there are lots of solid building blocks out there. But watch out for the cost structure!