That "HashiCorp" in the title is a farce. The "Hashi" of the HashiCorp is far gone. IBM and "Corp" are redundant.
Community Editions don't make sense to the MBAs at IBM unless killing those off would create a global backlash (loss of goodwill for no positive revenue outcome is not good to an MBA. If loss of goodwill however, results in positive revenue, that's a good thing to an MBA. This is a well regarded, well studied and highly respected, revenue boosting MBA concept called "enshittification"). MBAs know that anything that doesn't directly translate into either viral growth or revenue next quarter is a waste of investor dollars. That makes anything else a risk to their own careers.
Boundary is past its viral growth stage, so the MBA needs to see revenue next quarter. Boundary fails the test there, so off with it!
HashiCorp was an innovative company founded by two engineers that built a lot of foundational, transformational infra products. Then, in a growth frenzy based off the traction of these tools they raised tremendous, vast sums of money and subsequently ~lost control of their vision~ hired MBAs, and the company went down the drain as well, or as MBAs call it - got acquired.
What remains of those tools are a BSL that focuses on maintenance operations and feature requests that strictly ride along a highly scrutinized cost-benefit decision matrix of rent seeking behavior off IBM's contracts with organizations that had already embraced those tools.
Billions of tokens have been burned in creating PPTs and Excel sheets, and emails discussing them, about why the Boundary Community Edition should be archived and abandoned.
A year ago I spent ~3 weeks ripping out every piece of Infisical in my infra and replacing it with a mix of Keycloak, Conjur and Age/SOPS. They are ugly, hard to work with, setup and debug - I'm not proud of this monster but I won't have to deal with what's going to happen to Infisical.
That's why Kubernetes has gained a lot of traction, not because 99% of companies need the scalability of Kubernetes, but it is a well understood concept, suite of services, and the vocabulary is very common now and well understood.
Nomad is a great fit for most compared to Kubernetes but failed to gain adoption because of these reasons.