Based on Steve's earlier comments, he considers Ion or MPD 'late game' techs in his games. So anything beyond that is probably not properly balanced.
Lately Steve's AAR games have started with ion tech and usually gotten into MPD before he moved on, so I'd be hesitant to call those "late game" although of course the man himself can always answer that.
Generally the tricky thing with missiles is that they are not symmetrical as tech level increases. What I mean is that they only have two techs that significantly improve their ability to dodge AMMs/PD, which are the speed (actually an engine tech, not a missile tech) and ECM. Meanwhile both AMMs and Gauss PD have three techs that can be used to significantly improve their interception ability: ECCM for both, and speed + agility for AMMs while beam PDs have both tracking speed and missile tracking bonus (Gauss also has ROF increases which compounds this). Thus early anti-missile tech is usually quite poor, while late-game missile defenses are easily impregnable except against the cheesiest of strategies.
It is similar to the famous linear vs quadratic power curves in e.g. Dungeons and Dragons when comparing Fighter and Wizard class power levels in concept.
The flip side is that as we reach the late game this may be intentional and it becomes more desirable for a fleet to have a more diverse weapons loadout as they can develop the technologies to do so, and missiles become just part of a fleet doctrine compared to a single-weapon early game doctrine. At this point the major issue is perhaps more attributable to the AI as others have stated which is not optimized for this kind of late-game diversification of doctrine and designs. Thus as a player it becomes kind of silly to bother optimizing one's late-game doctrines to such an extent aside from RP reasons (and RP of course is always aided by a credible non-player actor or three).
