rather than relying on homogenous building block
No different to AWS or Azure. As upgraded servers become available, they don't just bin what's left.
Google's cloud thought it was out of memory for a couple of hours last week. The outage wasn't a biggie – it hit last Wednesday, June 7th, and meant that “7.7% of active applications on the Google App Engine service experienced severely elevated latency; requests that typically take under 500ms to serve were taking many …
That being made of common blocks suggests common behavior. Houses are made of common blocks too!. There will also be external factors local to the facility, it's scale, and likely rarely tested to extremes if workloads in all possible combination. Will be interesting if they give away a bit more St some point though.