Just to show how serious it is, I have worked at Google for many years and have seen code yellows a few times, but code red was always just a theoretical thing.
I think it's great that the management woke up and understands that they need to disrupt themselves.
I also worked at Google. Code Reds happen when the management realizes they are signficantly behind (or being caught up to) from external sources and at a risk of an immediate loss of reputation. Often times, the engineers have already tried to self-organize and have built 2-3 not-ready-yet solutions in 120% time, all of which either got deprecated or aren't management's favorite product (I'm cynical having seen this cycle happen more than once).
Jeff Dean already said externally that Google is purposefully not launching great models in the way OpenAI and others are, for all the reasons laid out in the article (and more). But, it's clear they use these things inside their products- probably not enough, because all of Google's main products that depend on ML have issues that will only be surpassed when we have a combined search/language model with a Google-scale index. Or maybe somebody will be able to do it with significantly less reference information than Google.
I think it's great that the management woke up and understands that they need to disrupt themselves.