Here is a Difficulty 7 done in 1h11min36sec.
The usual Cloak/Webbed/Reju with 80 planets, in the Grid/Crosshatch.
I was stupidly lucky with that seed, given I got the Spire Archive literally next door. It was the planet north of my Homeworld. GCAs weren't as kind, but I was able to boost my forces quickly enough to go kill the AI. I picked a turbo raider and a Pulsar Tank thingy since it double health of the low-shield turbo raiders. I had others options for speedrunning, but it kept the AI mark at 2 so I don't know if it would have made things faster in the end.
I got three battlestations with turrets, and they were honestly carrying the day. Plus two overloader supports with engineers (I curse when they get no engineers). I think it might be tricky to lower the time needed with vanilla settings, as all this made for an exceptional seed. I followed Pireciter's direction and grabbed a couple Officer fleets: the Thanatos-zombifying one and the forcefield Ark. They were really good and definitively worth the investment. It's a big difference to my usual games where I find them rather useless, however playing quickly so I win within one+ hour rather than five hours in makes a lot of difference.
I found it necessary to science-hack a couple ARS, I got two decent vaults. Tech-wise it was all over the place. Subterfuge, Generalist, Raid and Fusion. I got a single metal upgrade not that quickly, but spend 500 science on every command station I had. This game I eschewed military ones more than in my last speedrun, as the 4x movespeed is overall more useful in multiple cases than the 50% damage boost of military.
As my screenshot show, it feels and seems best to trace a line toward your objectives rather than leave AI planets in the way. It makes reinforcing much, much, much faster. Mostly though, planets on the outer circles are often too tough for initial forces. It was better to go bottom, top, bottom, east for a couple picks, then all the way north to victory.
Surprisingly you don't need AIP reducers or GCA for the first 30min. My method was to leave 3 spiders, 1 Nucleo and 0-3 MLRS from my battlestation. At mark 5-7 they were enough to deter or kill AI stragglers and waves. I had a "guard" fleet that I swapped some ships that cost me metal or couldn't afford to mark up, which helped hold of waves in the 30 strength range. This was excessively efficient, since it meant I had potent turrets to help kill the AI forces on the planet and then to defend the planets, all for a minimal metal cost.
The key was to get to the zenith matter converter, in my opinion. It just gives too much metal to ignore, and doesn't need to be built first like Fallen Spire mining rig. I didn't pop any beacon this game, although it might have helped cut down on the time: 5000+2000 form Nano and Marauder would have helped... but it's also insanely risky since once you go for the AI Homeworld you're committed.
It was fun. A lot more fun than the slow approach of fighting difficulty9, or the anemic response of diff5. Diff7 kept me more on my toes, as I couldn't afford to leave my planets too undefended. It's weird being in situation where not building a second or third forcefield to protect a logistical is the superior choice, but it's pretty exhilarating: not needing to build every defense I can make on every planet.
The AI Homeworld was somewhat lucky. No tesla, just parasite... and spider guard posts. Not sentry spam either, so I was able to build turrets and expand.
Will try again!
***
(Note: Pirectier, do post screenshots of your wins. It makes it easier to picture things and get invested. Plus that's how we notice things that others players do or don't do.)
