Well, but it's not like AWS didn't have its share of downtime in the last couple of weeks. Back then the recommendation of the arm-chair-admins was to move to dedicated hardware you own. Oh the irony.
With the last round of failures some of the people affected reported that they couldn't start up instances in other datacenters since those were overloaded. Still, I agree - the underlying problem is cross-datacenter redundancy but that's a very difficult problem to tackle and might just not be economically feasible.
