The cloud is built with a reliable infrastructure in mind. We have separate zones within a region to protect against local faults. We also have multiple regions to protect our services from catastrophic events that affect a whole region. Depending on the service this failover will happen automatically or it will be up to you as the consumer to make sure you have this fault tolerance in place. The Service Level Agreement for the different services can be found at https://azure.microsoft.com/en-au/support/legal/sla/. If you run a certain service in multiple regions it will be much more reliable than running it in a single region.
The cloud also offers predictability in different forms. Performance predictability makes sure that you will have computing resources available when needed. Concepts such as auto-scaling, high availability, and load-balancing help achieve this. Cost predictability on the other hand helps you forecast the costs for your consumption of the resources you use by doing trend analysis.