There's several factors at play. First, Google simply does not have any ability to compete in the consumer gaming space because they don't understand it. Second, to make the project work required an enormous expenditure across hardware, software, deployment, and game studios. If the product wasn't absurdly successful, it would be a failure simply because its profit margins would be low. Third, there's no real way to make money doing machine learning on gamer behavior (the way this works in mobile, where many games including ads) so the profit margin would be low. Fourth, I saw a number of preeminent engineers who worked on the project leave shortly after it was launched, or some time later (when it first started becoming obvious the project wasn't a hit). Fifth, leadership pitched this as something that only Google could do, that Google's unique hardware and physical presence in POPs meant they would have significant advantages was obviously wrong (multiple companies always had the technical acumen and production infrastructure to make this happen).
What really blows me away is how close Google Cloud is becoming to something that Google would have to cancel because they can't get the profit margins to compare to ads.
I imagine GC is one of the big three cloud platforms (with AWS, Azure) so I am absolutely intrigued by your suggestion. How bad is it going, really? They've made a lot of investment in Firebase also. I could probably see them shake things up drastically, but shutting down sounds like suicide.
The technology isn't why it failed, Google is. Google was Google before it launched, and after it launched. So anyone familiar with Google could have predicted this failure before the actual launch.