What however seems to be true is that they indeed invested in optimizations that allow them a lower training cost (from what several people that criticized their research paper have told). Let's now get science play its role and just independently redo what's in the paper.

A "David Vs Goliath" story is enough to make all media outlets forget to think : the "5 millions for DS vs 50 billion for OpenAI" is wrong on all levels (they don't measure the same things).

I read this very interesting post by @ghuntley.com a few months ago on VScode and why the Visual Studio Code ecosystem is designed to fracture

ghuntley.com/fracture/