Screw this. I'm warping into this discussion.
DANGER: WALL OF TEXT
See, my understanding of technology advancement is somewhat like this: You can only go so far with one piece of tech until it stops.
Take a bow, for instance. The key concept is that you nock an arrow, pull the string, and let the arrow fly. You can advance the bow's technology -- give it better sights, help you hold the arrow easier, use better composite materials -- but there's a certain point where you've made the best bow statistically possible, and as much as you'd like to try, the bow cannot get any better. The arrow can follow the similar philosophy.
Get what I mean? You'll at some point just end up with the best possible and there will be no further advancement. Sure, you'll have a bow you pull back, it auto-locks for, idunno, the nearest cruiser, shoots the arrow that pierces through the reactor...but wouldn't a railgun be better?
Oh look. Now we need to test the railgun. So we go back to the physics books, look through what we know, exploit those rules in physics, and create something new. So we make a railgun, and so it goes and we push the technological frontier onwards...
And so here we arrive at the fatal flaw of AI technology. Or rather, just any one entity building all the weapons, like Yamatai's "I request new weapon. AI develops new weapon. I am happy." It's a psychological thing -- just like how artists have art styles, writers have writing styles, and how sci-fi races tend to build things a certain way, such as Yamatai's sharp edges, Elysia's smooth curves, and USO's...I'm not sure because I don't follow the plot (please don't kill me).
One entity building weapons will result in all weapons following the rules and design patterns that one entity prefers. For instance, say a weapons development AI was wired to favor blunt weapon damage and kinetic force. That one entity would love to slap railguns on everything, and make big power hammers. If, for instance, you want something like plasma and cutting edges, that one entity will probably say something like "Yeah, but plasma can just be magnetically shielded away and cutting edges can't deliver force through armor." Even though all of us know that both have their own benefits.
There we go. That's AI developing tech covered. Onwards to the actual power creep.
If we want SARP to be realistic in its power creep, I personally feel that we need each faction to feel somewhat different in their design philosophies. For instance, if Yamatai, the faction which I personally associated with fast, heavily-armed gunships and modular PAs suddenly came up with something like the Hostile Power Armor, a solid metal block of a power armor, that would be weird. Unless Yamatai got a Neppy PA designer to make something and that guy decided, "Hey! I'm gonna take all this Yama-Dura, shove servos and circuits in it, and call it an armor!"
Pretty fat chance of that happening.
We need every faction's weapons to feel like their own. Sure, Elysia and Yamatai can trade notes and base designs off of each other (Tethys is #1 Elysian PA <3) but we need more than just every faction to prefer a certain type of shape.
Well, what do I mean by that? I refer back to the design entities. One prefers blunt trauma and kinetic damage, the other prefers cutting edges and plasma. Both have their own benefits (and drawbacks) but both work differently, feel different, and all-in-all are different. Yamatai to me are the kings of aerospace technology, with their aether. Neplesia is amazing at brawling with their heavy armors, and anything big and punchy.
And so it goes, every race preferring some form of design, weaponry, and style. That is my personal solution to the "power creep" problem. Sure, some faction can prefer shielding and such, and will be pretty hard to beat by a faction which doesn't like plasma, but I refer to the theory of trading notes, where the faction that doesn't like plasma tries its best to replicate the idea of plasma weaponry and makes something effective
enough to ward them off, but by no means wins wars.