My takeaway from that is very early startups should be in person, but beyond that very early stage remote will make sense especially as devs can be picky and want to work remote.
And mine is that you can't tell much from that anecdote. The failing startup might be failing for any number of reasons. There are counterpoints of fully remote startups that have worked out well. So basically, it's just noise. There isn't even an attempt made at explaining why is working remotely the problem, they might not be suited to it, or that they tried before being ready for it as OP stated.
Yes, In Person is always better where there is more "chaos", but for established companies where most people have "tasks" to do, remote works just fine.