Re: 1 million users?
1. That's what their initial US license is for, which is more than enough to cover their beta phase.
2. Most of those satellites aren't over the US at any given moment. However, they don't have the spectrum licensing for Rest-of-World yet, so any given sat is basically doing nothing for 80% of the time. Their licensing is for 1million US users, necessarily restricted to using a small subset of the 1100 sats currently in orbit because of orbital mechanics.
3. Starlink is never going to serve hundreds of millions of Americans because the majority of Americans live in major metro areas and have acceptable(ish) internet access. If 1% of New Yorkers tried to jump on StarLink, the radio congestion on satellites with LoS to NYC would be... ugly. Albeit there's obviously scope for multi-tenant buildings to share a single antennae depending on how they structure their sales and if they provide MSP/ISP support rather strictly selling direct-to-consumer on a one-dish-one-customer basis.
4. Spectrum licenses are pending in many countries, including the UK/OfCom, at which point they can increase utilisation of their network. But having spectrum is only one step - you then need ground stations configured at IXPs, bit barns, with transit providers, etc to actually hand user data off to. This is a lot like building out Tesla's SuperCharger network. They need to bootstrap the infrastructure before the product becomes useful. Additionally, the generation of satellites launched so far don't have the inter-sat links (aside from some on the very latest launch), which means terminals in the ocean or rural Africa could connect to StarLink but there would be nowhere for the satellites to send the data - no ground station, and no ability to relay to a sats which can see a ground link.