They didn't evolve into SF; SF was a project inside of VA that eventually became the flagship of what remained after the hardware and related services were excised. When they started (1998/99), Git wasn't a viable option (the first version of it wasn't released until 2005, by which point SF had ballooned to an enormous scale at the time, with it's own product inertia, and it would be a few more years before Git would become a major VCS itself, which is when Github started, and by then VA/SF was in decline and had changed hands several times.
> it would be a few more years before Git would become a major VCS itself
People forget that in the olden days we used Subversion and Bazaar (well, the latter if you were Canonical-adjacent), and before that CVS.
And before that, SCCS.
Going back decades, it's all people going "this sucks, I'm writing my own VCS", and for whatever reason Git was the one that gained traction in that particularly sticky and slippery swamp.
I'd love to know what the "thinking" was behind getting rid of the hardware business. We bought some Penguin Computing servers after VA left the market.
1. I have some serious biases against Penguin at the time (for...reasons) and frankly was never impressed with their product.
2. Without getting into a bunch of weird minutiae, we got big enough to be a threat to people who could afford to bleed us; case in point, we started doing incredibly well in HPC clusters, and big vendors like IBM and Dell started to offer severely below cost hardware packaged with their full services that completely undercut a business that already had thin margins.
IMHO, we made better gear at the time, but we were not in a market as wide and deep for linux optimized machines as it is now. It's not an unusual story in the valley. We did have a deep talent bench that ended up in key roles in a bunch of firms that are doing well: Google, Apple, et al.
(disclosure: I was on the "Ignition team" for SF)