Three things displeased me. First, one of the (many) ways in which TA was revolutionary for its time was that it largely emphasized thinking over rapid micromanagement and SC was a regression in this regard, an unnecessary concession to the RTS scene of the day. Second, how technically underwhelming it was in terms of pushing the envelope in scale, unit counts etc. compared to the gigantic leap in available performance between 1997 and 2007. Third, the unsolvable technical limitation that any game against the AI (including custom AIs) would eventually crawl to an unplayable lagfest - this is still true even on today's hardware.
Well, the primary problem was that Supreme Commander made a bet that "Teh Pretty(tm)" was worth pursuing (probably as a result of complaints about TA). SupCom absolutely murdered graphics cards of the day--and it still plays like a dog on modern integrated graphics. Consequently, unit count was always going to be limited.
I suspect that modern graphics cards could probably chew through this--but you need to build an engine that leans into the way modern graphics cards work.
Emphasis on graphical fidelity was an issue when it came out, but the long term performance nightmares are all single threaded CPU bottlenecks. The whole engine just scaled really poorly with increased unit counts and AI/pathfinding computations. For a game that's basically all about scaling that was something of a dealbreaker.
