On the surface I agree with you. In practice, I've gone the other direction and have become more pro-open-source over the years.
One example is MATLAB. I like MATLAB, and consider myself fairly good at it. People come to me to ask MATLAB questions. With that said, my company has floating licenses and these are a pain. Mathworks is very responsive in their customer service, but when you find a bug, you have to work around it or wait until they fix it. On the odd occasion where you want to actually distribute a script, you need to (maybe?) have the end-user download and install the (free as in beer) runtime separately.
I've switched the vast majority of my data analysis and other scripts to Python, and I no longer have to search for co-workers who left their copy of MATLAB open. When I find a bug, I can actually fix it myself and even return the fix to the module's project, along with any other feature that I find to be missing. When I need to distribute a script, I just make sure that I'm not using some forbidden-fruit GPL module (the ecosystem is mostly BSD) and zip the whole shebang up with one of the py-to-exe tools without consulting the frigging lawyer.
It's not all-rosy, for sure. Scientific/technical computing on Python has a higher learning curve than MATLAB. While vast help exists for Python trouble, MATLAB has all of the help concentrated in one place which makes finding solutions easier. One unexpected benefit to Python is the GUI. At first blush, MATLAB holds the high ground with its GUIDE visual GUI builder. But for anything more than a few simple controls, GUIDE is an unholy beast to work with. I've found my life much better with Python and it's wide choice of GUI frameworks. Even setting up the whole GUI with a text editor in Tkinter is worth the up-front time investment vs. the misleading initial ease of using GUIDE.
On the topic of SAS, one product that I do use of theirs is JMP. I have to admit it is faster (for me) for quick-and-dirty data analysis than using Python. I think I'd like to code up a Python application to do some of my most common JMP workflows... not try to reimplement the whole thing.