I think your probability calculation might be a bit off. The math doesn't go through.
I should say ahead of time, I don't know much about these 4-9s vs 5-9s. I interpret them as probability of not failing. IE, 4-9, means 99.99%, which means the probability of failure is .0001. If that's wrong, the rest of this doesn't work out.
Lets try different numbers. Choice A has a probability of 25% of failing, Choice B has a probability of 1% of failing.
How many A do we need such that the probability of them all failing is less than 1%?
If I have 2xA, what is the probability that they both fail (assuming they are independent)?
P(A1) and P(A2) = .25 * .25 = .0625 (6%)
What if we add a third:
.25 * .25 * .25 = .015625 (1.2%)
And a fourth
.0039 (.39%)
So, 4 of these 25% data centers is better than a single 1% data center.
The case is even stronger for the 4-9s vs 5-9s example.
4-9s (if I understand) means 99.99%. Or, .01% of failure (P=.0001).
5--9s means 99.999%, or .001% of failure (P=.00001).
2 x 4-9s is .0001 * .0001 = 0.00000001 , which is 0.000001%, which is 99.9999 (6-9s).
To me, it makes perfect sense to do the "google" thing. This is exactly the reason that they fill their data centers with low-cost commodity hardware instead of high cost servers.