and remove all launches where SpaceX was the customer (a net loss unless the payload they launch has payoff), and specifically look at the GTO launches (the ones that pay the most money). There are fewer and fewer GTO launches, and the number of customers has been going down. A cursory glance shows 2017 had 8, 8 in 2018, 4 in 2019, 2 in 2020, and about 8 planned in 2021 (we'll see, these typically slip quite a bit. The point being that they need profitable launches (not a small satellite they rideshare with 10 starlinks on) to sustain a launch business with that many engineers on it. I think 2021 will launch maybe half the GTO/GEO they're targeting this year due to delays, as was the case in 2020.
Regardless, commercial launches is what I was talking about as well. Do you have a reference for the idea that GTO launches are the only ones which are profitable / significantly more profitable than other launches? I haven't seen that. Because otherwise it seems pretty out there to only look at a small subset of their commercial launches.
And I'm assuming they'll do more GTO launches once Starship + Superheavy is online.
It's a 6x difference in cost per kg.
Also are we looking at the same chart? Not sure how you can make the claim that the number of launches has been dropping every year... every year since when? 2020 was the most launches ever, and in 2021 they are expected to do more commercial customer launches than all launches in 2020 + Starlink launches on top of that.
https://en.wikipedia.org/wiki/Falcon_9#Launch_outcomes