Google is heavily investing in cloud; in 2015 they embarked on a sizable shift, where they decided it was worth using their considerable head starts in databases and machine learning.

They open-sourced Kubernetes and Tensorflow that year, both with a view toward cannibalizing AWS's lead. They released Apache Beam the following year, but it hasn't caught on the same way.

At this point, Kubernetes is winning everywhere, so why not use Google's infrastructure for it? , and tensorflow has enough adoption that you might go to Google Cloud just because they integrate so tightly with it.

More recently, Google purchased Kaggle, and has turned Jupyter notebooks into Colab, as well as hiring Jake VanDerPlas out of UW. They're trying to turn GCP as the easy, first choice solution for running and versioning your darn notebooks. But if you're already there, why not use BigTable to pull the data INTO the notebooks...?

There are definitely tone-deaf steps they're taking, but Google is building serious advantages into their cloud offerings.