if we want to know that space has "arrived" as a new technology or "thing", what we need to see is a breakthrough in the cost to manufacture a unit of capability
That’s exactly right. The big hurdles when it comes to reducing the cost of space assets is that 1) they HAVE to work and 2) they're rarely built in volume. Those factors mean that even a bare minimum product is going to cost a LOT of money. And...if you're already spending a lot of money on a base model, you might as well spend more for the P-D, as it were. Intuitively, volume enables amortization of capital to reduce your average cost, but--at least today--there's still a very strong industry bent to #1 from above, so there's still a pretty high asymptote on cost. You really need to get into HUGE volumes (like starlink) to drive average cost way down, but of course you need many billions of dollars to get something like that off the ground...so you're kind of just trading one hurdle for another.
Don't get me wrong--it will happen. It just won't happen quickly. The tipping point really is when someone finally turns on a fully armed and operational internet constellation. As you would imagine there's already folks wanting to use the platform(s) from the Big 3 (Starlink, Oneweb, Telesat) for all manner of applications, because the base cost of those platform is SOOO much lower than a legacy solution. The catch 22 is that the follow-ons all rely on the big 3 actually producing a commonized, low-cost, payload-less platform, because the internet constellation volume is explicitly the cost enabler. All of those platforms would cost [more or less] the same as a legacy Space 1.0 solution if they were only built in legacy volumes. Oh, and everyone working those constellations has little time for some commonized variant of their platform, so all the ambulance chasers kind of have to wait.
The launch cost improvement surely helps but it's not the long pole in the tent. Or at the very least, it's no longer the long pole in the tent.
Right again. While reducing the cost of a launch by tens of millions of dollars is undeniably huge--thanks solely to SpaceX--the total capital spent putting an asset (or rocket full of stuff) on orbit is at a minimum ~3x the cost of a SpaceX launch. For GEOcomms its probably more like a minimum of 5-6x, and for government stuff it can easily be 10x or more. Then the business model around those assets (or however you want to quantify "value" to the government...) is at a minimum another 2-3x multiplier. So...from a relative sense, the money saved on a SpaceX launch is a fairly small piece of the pie.
Again, its definitely non-zero. If you're a for-profit company, tens of millions of dollars means something. But...in the big picture it's not necessarily everything.