IBM announced it was buying HashiCorp last month, and my first reaction was not strategic. It was to open the directory where my Terraform lives and look at it the way you look at a car you have just learned the manufacturer has been taken over: same vehicle, slightly different feeling about who is responsible for it now.
The deal itself has been picked over plenty, and I have no special insight into IBM's intentions. What I do have is a pile of infrastructure described in HashiCorp Configuration Language, a tool I genuinely rely on, and a licence situation that changed under my feet last year and now sits inside a company with a long memory and a very particular relationship with the products it acquires. So this is less a hot take on the acquisition and more an honest audit of what it means for the actual tool on my actual disk.
The licence was already the real story
It is easy to treat the acquisition as the event, but for anyone using Terraform the licence change came first and matters more. HashiCorp moved Terraform and most of its other products from the Mozilla Public Licence to the Business Source Licence in August 2023. For me, running it to manage my own homelab and a handful of personal projects, the BSL changes nothing in practice: I am not building a competing managed service, which is the thing the BSL is written to prevent. My usage was fine before the change and it is fine after it.
What the licence change did do was crystallise something I had been lazy about, which is that I had quietly assumed Terraform would always be open in the sense I cared about, and that assumption was never actually load-bearing. The OpenTofu fork happened precisely because enough people felt the same way and had more at stake than I do. I have been watching it with interest rather than urgency.
What I am actually going to do
Very little, immediately, which is the boring and correct answer. The acquisition does not change the licence, it does not break my existing configurations, and it does not suddenly make a fork I have not adopted any more or less necessary than it was a month ago. Reacting to corporate news by ripping up working infrastructure is how you turn someone else's press release into your own outage.
What it has done is move OpenTofu up my list from "interesting" to "I should run my configs through it and see". The whole appeal of a drop-in fork is that the cost of evaluating it is low: point it at the same configuration, see whether it plans cleanly, and now I have a tested escape hatch I did not have before. I do not have to switch. I just want to know that I could, cheaply, if the day comes when the stewardship of the original stops suiting me. That optionality is worth an afternoon.
The other thing I am doing is the thing I should have been doing anyway, which is making sure I am not married to any single tool's quirks more than I have to be. State is state, providers are providers, and the more of my setup that would survive a tool swap, the less any acquisition or licence change can actually do to me.
The wider feeling
There is a slightly melancholy thread to all of this. A lot of the infrastructure tooling I grew up on was built by independent companies with strong opinions, and one by one they are becoming line items in much larger ones. That is not a scandal, it is just what happens, and IBM is not a villain for buying a company any more than the company was for selling. But it is a reminder that the tools we lean on are products owned by someone, and ownership changes, and the only real defence is to lean in ways that you could unlean if you had to.
So I will keep using Terraform, because it works and the acquisition has not changed that. And I will spend an afternoon making sure I am not trapped, because the acquisition is a useful reminder that I never quite checked. That is the whole of my reaction: no drama, one tested escape hatch, and a slightly more honest relationship with a tool I had been taking for granted.