SpaceX is developing the launch capability so this is even more concerning for Loon since they can't really compete on that reduced cost. Launch is also mostly an up front cost, balloon launch, tracking, and recovery is recurring.
I think you're right on balloon concentration and there's a narrow use case here for things like congested events, but it's pretty niche. I think Starlink is trying to get uniform satellite distribution? Global coverage in the common case seems a lot more important than concentration as long as you can meet the minimum bandwidth required (which they probably can).
I think servicing is a good point, maybe the balloons can improve faster as a result, but it's also a negative (having to service/track them all of the time). It's also not global coverage - has Loon given up on this goal entirely?
Global coverage for Starlink means access to a global customer market which can help cover their increased costs (since I'd guess margins after launch are low). Without a global market even if Loon has lower costs it'll be harder for them to get the revenue to cover them.
> "Google already has experience with being an ISP with fiber and fi."
They abandoned Fiber and I hear Fi kind of sucks, but you're right that this is a point in their favor.
These feel like rationalizations of a losing position to me.
If I worked at Loon I would be worried.