Google Cloud has issued a security advisory regarding three side-channel vulnerabilities identified in AMD Turin processors. These flaws present a risk of unauthorized data exposure across processes running within the same virtual machine (VM). While Google has already implemented infrastructure-level mitigations to prevent cross-VM and VM-to-host data leakage, users are now required to take action at the guest operating system level to secure their individual workloads.
The vulnerabilities function as side-channel attacks, a class of flaw that typically exploits the microarchitectural implementation of processors rather than direct software bugs. In this instance, the flaws allow an attacker capable of executing code on a specific processor core to infer sensitive data from other processes or the kernel operating on that same core. By monitoring timing variations or other subtle hardware signals, an attacker may potentially reconstruct information that should remain isolated within the memory space of other processes.
The affected hardware is the AMD Turin CPU architecture, which is currently deployed within Google Cloud’s compute environment. Because these vulnerabilities reside at the hardware level, they affect the fundamental way the processor handles instruction execution and data isolation. When such flaws are discovered, the standard industry response involves a combination of microcode updates and operating system-level patches that enforce stricter boundaries between processes and prevent the leakage of information through shared hardware resources.
To fully remediate the risk, Google Cloud has instructed customers to apply the necessary patches to their guest operating systems. Following the application of these updates, a stop-and-start of the affected VMs is required to ensure the new security configurations are fully initialized and active. Failure to perform this restart may leave the VM environment in a state where the underlying hardware remains susceptible to the identified side-channel techniques.
This incident highlights the ongoing challenge of maintaining isolation in multi-tenant cloud environments where hardware performance optimizations can inadvertently create security gaps. As processor architectures become increasingly complex to achieve higher throughput, the surface area for side-channel analysis continues to grow. These types of vulnerabilities underscore the necessity for a layered defense strategy, where infrastructure providers manage the platform-level security while users remain responsible for maintaining the integrity of their own guest environments.





