For instance, Facebook can open source React because it's something they need, but it would not be a viable product on its own. Releasing it as open source doesn't give them any disadvantage. Similarly, a GPU manufacturer can open source their drivers because this won't prevent them from selling their hardware (and in theory should allow them to sell more hardware).

In theory, I'd expect a printer manufacturer to be able to open source their software, because their product is the printer not the software. In fact, I don't get why open sourcing 3D printer software should be a disadvantage, and the article doesn't explain this.

Meta does OS well for their scale. React, Llama, PyTorch come to mind.

For MS: Typescript, vscode, .net come to mind.

- developing features or bug fixes in the open source product or an extension that are needed now by the customer

- developing a custom project (not in the open, or actually sometimes in the open) on top of your open source product for the customer

- supporting the customer for some integration with your open source product.

You can also try to sell paid extensions (that are still open source). That would be the case for XWiki Pro apps, or many WordPress extensions.

There's another way to do it, is develop your open source and sell your developer expertise for unrelated customer projects. I've worked somewhere working like this, but don't really believe it works well because you are not actually spending time in your project when you are doing this, and you are not incentivized to develop your own product when you do this. You are basically a consulting company. But you can still decide to do a bit of this.

Consulting and support can work for solo developers, but that might push quite some pressure and make it difficult to disconnect depending on the way you do this.