The issue is not companies but governance. OSS licenses and companies are fine. Companies have a natural conflict of interest that can lead them to take software projects they control in a direction that suits their revenue goals but not necessarily the needs/wants of its users. That happens over and over again. It's their nature. This can means changes in direction/focus or worst case license changes that limit what you can do.

The solution is having proper governance for OSS projects that matter with independent organizations made up of developers, companies, and users taking care of the governance. A lot of projects that have that have last for decades and will likely survive for decades more.

And part of that solution is to also steer clear of projects without that. I've been burned a couple of times now getting stuck with OSS components where the license was changed and the companies behind it had their little IPOs and started serving share holders instead of users (elastic, redis, mongo, etc). I only briefly used Mongo and I got a whiff of where things were going and just cut loose from it. With Elastic the license shenenigans started shortly after their IPO and things have been very disruptive to the community (with half using Opensearch now). With Redis I planned the switch to Valkey the second it was announced. Clear cut case of cutting loose. Valkey looks like it has proper governance. Redis never had that.

Ollama seems relatively OK by this benchmark. The software (ollama server) is MIT licensed and there appears to be no contributor license agreement in place. But it's a small group of people that do most of the coding and they all work for the same vc funded company behind ollama. That's not proper governance. They could fail. They could relicense. They could decide that they don't like open source after all. Etc. Worth considering before you bet your company on making this a foundational piece of your tech stack.