Google Colab has a very very strong Engineering team. Part of Google Research they definetly changed the game when they offer a free product which 1) increased collaboration across ML research 2) offer everybody access to GPUs first K80 and now T4 which is great among researchers and students. Now many other notebooks products now want to be the Enterprise colab version
Very good question. They target different audience,
Google has many notebooks solutions: colab (free) colab enterprise (monthly) kaggle, ai platform notebooks and datalab. Colab is targeted to students/researchers which are just experimenting, as there is no guarantee that kernel will run more than 24hrs. The paid version removes this restriction. AI notebooks is targeted for enterprise data scientists (Jupyterlab) which require patches/VPC-SC/IAM integration, security, etc. Datalab is the very first version of the notebooks and that's going away. Kaggle is mainly for competitions. In the end Google will have only 2 Notebooks colab and AIP notebooks
I dunno, I believe Microsoft's Azure Notebooks predates Colab and perhaps gave Google the kick to make it public, and Colab was a spinout of an internal tool with a rather rocky relationship with open source that has still been a pain point I run into using it for teaching ("what is the difference between Colab and Jupyter?" is a question without an especially clear answer). Also they stole the name from GE's old internal social network.
That latter one, at least, was not entirely serious criticism.
I'm not sure that even Colab is an unmitigated win for the company, it does definitely have "that AWS feeling" of an internal tool that was made a product without really retooling it for that purpose thoroughly. This has kind of become the norm in cloud platforms, though.
