My feeds have been one long primal scream for a fortnight. Microsoft announced it was buying GitHub for somewhere around seven and a half billion dollars, the deal got confirmed in early June, and the internet promptly lost its composure. Half of it is "embrace, extend, extinguish" energy; the other half is people quietly migrating to GitLab to make a point. I've watched the whole thing with a cup of tea and a slightly raised eyebrow.
Here's the thing I keep coming back to. Git is distributed. That is the entire point of it. Every clone on every laptop is a full copy of the history. If GitHub vanished tomorrow, or turned into something I disliked, the cost of leaving is git remote set-url and a bit of CI rewiring. The pull requests and issues are the sticky bit, not the code, and that's a real concern, but it's not the apocalypse people are billing it as.
What's actually interesting is the shape of the Microsoft that's doing the buying. This isn't 2001 Microsoft. This is the company that open-sourced .NET, ships a half-decent Linux subsystem in Windows, and put Visual Studio Code in front of half the developers I know, including me. They didn't buy GitHub to kill it. They bought it because developers are the kingmakers now, and GitHub is where the developers live. You don't pay seven billion for something you intend to set fire to.
Do I think it'll be fine forever? No. Acquisitions have a way of slowly sanding the soul off a thing you liked, and there'll be a quarter three years from now where some metric needs juicing and the experience gets a little worse. That's just gravity. But the GitLab stampede feels like a reaction to the logo more than the substance. I keep an account over there, I mirror a couple of things, and I'll watch how this plays out. If it goes bad I'll leave, calmly, with my history intact, because that's what git is for.
The part I'd actually keep an eye on isn't the hosting at all, it's GitHub's quieter dependencies. The npm ecosystem leans on it, countless CI pipelines clone from it without a second thought, and a generation of developers treats "push to GitHub" and "the project exists" as the same sentence. That concentration was a risk before Microsoft showed up and it's the same risk now, just with a different owner. If anything the acquisition is a useful prompt to remember that you should never have all your eggs in one provider's basket, whoever owns the basket this week.
So my plan is boring. Keep a mirror somewhere else, make sure my CI can be repointed without a rewrite, and otherwise carry on. The repos that matter to me get pushed to two remotes already, a habit I picked up years ago for exactly this sort of "what if the host changes its mind" reason. Turns out the right answer to a big acquisition is the same as the right answer to a host going down at 3am: have somewhere else to go, and don't make the move under pressure.
For now I'm not migrating anything in a panic. I've moved enough infrastructure under deadline pressure to know that the worst decisions get made in the first fortnight of an announcement, when nobody has any new information and everyone has a lot of feelings. Give it a year. Then we'll have something to actually judge.