The biggest failure with scale I have seen in my own personal experience is fragmentation. The business knows they want to grow or diversify or whatever. They know what their desired end state is, and they kind of (at least they think they do) know what needs to happen to get there. This is wildly complicated with too many unknown unknowns so they backwards plan and short circuit until they reach their desired end state.

The end result is a bunch of teams each doing their own things in their own way and each with their own overhead. It is both expensive, high risk, and fragile. As developers we call this tech debt. If you think refactoring technology is hard then you are in for a real treat, because refactoring technology is child's play compared to refactoring people.

Of course the flip side to that is micro-management from the top. Yes, there is a centralized design, but nothing gets done because the micro-management gets stretched too thin to accommodate real world variability that comes with scaling in different directions.

I have never seen anybody solve for this in technology organizations as well as the Army. The Army has a vaguely new (15-20 years old) organizational philosophy called Mission Command. It is complicated, but the summary of it is execute to your leader's intent. The idea is that each subordinate leader knows what the boss wants and they own what constraints are imposed upon them. Just don't violate the constraints imposed upon you, but otherwise do whatever you think best to accomplish the boss's intent. The imposed constraints provide known limitations to prevent unintentional fragmentation but otherwise the subordinate leader is given maximum flexibility to move quickly and make decisions as operations dictate. Then once the initial boss's intentions are accomplished organizational refactoring occurs at lower cost.