Never trust your business, applications, or critical data to a cloud service because you are at the mercy of the provider both for security and availability, neither of which are terribly likely. Cloud services are the .coms of the 2nd decade of the 21st century, they come and go and with them so go your data and possibly your entire enterprise. Typically the argument is that larger brands are safer, that a company like Google would not wipe out a service leaving their customers or partners high and dry, that they would be safe.
That would be a false assumption.
It is necessary to understand the mathematics of serial risk to evaluate the risk-weighted cost of integrating a cloud-provisioned service into a business. It is important to note that this is entirely different from integrating third party code, which just as frequently becomes abandonware; while abandonware can result in substantial enterprise costs in engineering an internally developed replacement it continues to function, a cloud service simply vanishes when the provisioning company “pivots” or craters, instantly breaking all dependent applications and even entire dependent enterprises: it is a zero day catastrophe.
Serial risks create an exponential risk of failure. When one establishes a business with N critical partners, the business risk of failure is mathematically similar to RAID 0. If each business has a probability of failure of X%, the chances of the business failing is 1-(1-X/100)^N. If X is 30% and your startup is dependent on another startup providing, say, a novel authentication mechanism to validate your cloud service, then the chances of failure for your startup rise from 30% to 51%. Two such dependencies and chances of failure rise to 64% (survival is a dismal 36%).