I am not mistaken, new technologies should generate more problems than proven ones. GCE are new but they are based on proven Google technologies stack.
Hardware, as reliable and optimized as they are, will likely die one day or another. Software technologies on top of these hardware make the real difference. Software Defined Network, Software Defined storage, Software Defined Data center, optimized and distributed cluster management, optimized workload management are essential for high availability , manageability, scalability and self healing of your data center and your applications with a minimum outage.
You can deploy your application and let's GCE to manage high availability and scalability of your applications( i.e. Snapshat) , however you can spend time to design your applications for high availability and scalability on top AWS ( Netflix), GCE or Azure.
Last year, Google Drive achieved 99,985% availability ( less than 90 minutes per year), You can not say the same for One drive.
In 2012, Gmail achieved 99.983% availability, we cant say the same for outlook.com
which experienced not so long 3 days outage.
Google Apps is more reliable than Office 365, etc..
Conclusion, show me how you manage your own applications ( Gmail, Search, Google Apps, Google Drive, etc..) on your own cloud computing engines before selling them to me.