Cloud computing has become indispensable in today’s computer landscape. The flexibility it offers for customers as well as for providers has become a crucial factor for large parts of the computer industry. Virtualization is the key technology that allows for sharing of hardware resources among different customers. The controlling software component, called hypervisor, provides a virtualized view of the computer resources and ensures separation of different guest virtual machines. However, this important cornerstone of cloud computing is not necessarily trustworthy. To mitigate this threat AMD introduced Secure Encrypted Virtualization, short SEV. SEV is a processor extension that encrypts guest memory in order to prevent a potentially malicious hypervisor from accessing guest data. In this paper we analyse whether the proposed features can resist a malicious hypervisor and discuss the trade-offs imposed by additional protection mechanisms. To do so, we developed a model of SEV’s security capabilities based on the available documentation as actual silicon implementations are not yet on the market. We found that the currently proposed version of SEV is not up to the task owing to three design shortcomings. First, as with standard AMD-V, under SEV, the virtual machine control block is not encrypted and handled directly by the hypervisor, allowing him to bypass VM memory encryption by executing conveniently chosen gadgets. Secondly the general purpose registers are not encrypted upon vmexit, leaking potentially sensitive data. Finally, the control of the nested pagetables allows a malicious hypervisor to closely control the execution of a VM and attack it with memory replay attacks.