Fully homomorphic encryption is a powerful idea and undoubtedly has a role to play in cloud computing. In principle, it means that the cloud provider can run the correspondent of any program the client wishes, while not obtaining access either to the argument data or the result data. When fully homomorphic encryption was first created in 2009, it was held up as a magic solution that would solve the big problem of cloud computing security (IBM). Unfortunately, however, this promise is unlikely to be realised. There are two fundamental reasons that appear to limit the contribution that fully homomorphic encryption can make to cloud security.