Perhaps I've misunderstood your point, but I can't see how the...

Perhaps I've misunderstood your point, but I can't see how the Appen situation has any relevance here. Appen's core problem was customer concentration; I don't have exact figures but from memory >80% of their revenue came from just five big customers. Those customers effectively used Appen's data-labelling AI functions to teach their own AI machines which then made Appen's ongoing services (glorified labour hire) largely redundant. Even after trying to diversify their customer base, when Google recently ended their contract it was a 30% hit to Appen's FY23 revenues.

Megaport are in a very different situation to this; looking at the annual report they have 2,637 customer 'logos', meaning approx. $77K of ARR per customer. The competitive advantage is twofold - network size (868 enabled DC's in 162 cities) and the speed/ease with which they can connect customers. On the second point, there's an interview on Spotify 'AIs Wide Open Podcast S2 E9' where from 15:20 onwards Reid explains the clear benefit with using an enabler like Megaport versus the tradition route of telco's in country. In this game scale matters and can in itself become the competitive advantage. That's my take but happy to hear if/how this is wrong.