As organizations expand their use of AI technologies and cloud workflows, data privacy stakes are also increasing — especially where multiple teams get involved in joint data analysis and machine learning (ML) model training.
Confidential Computing is a powerful cloud security model that protects sensitive data where it is most vulnerable. Unlike typical encryption methods that protect data at rest and in transit, it secures data during processing by running workloads in trusted execution environments. Doing so creates isolated, encrypted environments within and between the CPU and its attached memory. This capability enables data to remain secure even while actively in use.
Google Cloud’s new C4D general purpose virtual machine (VM) offering — which integrates its highly optimized Titanium infrastructure layer with AMD’s 5th Gen EPYC processors (code-named Turin) — supports Confidential Computing via AMD’s Secure Encrypted Virtualization (SEV).
SEV encrypts the memory of individual VMs with unique keys managed by the processor itself, ensuring that data remains confidential and isolated even from the host system and hypervisor. This capability is enabled by Infinity Guard technology, which is integrated across multiple generations of AMD’s EPYC processors.
The combination of Confidential Computing and 5th Gen EPYC delivers higher performance and better efficiency for demanding workloads that use encryption, allowing enterprises to run workloads with confidential data at scale without significant performance trade-offs.
These advancements can further fuel use cases that rely on both security and high compute power, including:
- Multi-party data collaboration (Confidential Space). Multiple teams across different organizations or departments — such as joint research among healthcare providers — can securely share and process sensitive data without exposing raw data to each other. This facilitates collaborative analytics and insights while maintaining strict data privacy.
- Privacy-preserving ML. Organizations can perform AI inference within a secure enclave, ensuring that proprietary or sensitive data remains protected throughout the process. This boosts trust when handling regulated or confidential information.
- Secure financial modeling and risk analysis. Financial institutions can perform complex simulations and risk assessments on encrypted data, helping to meet regulatory compliance without exposing customer information or trade secrets.
- Confidential cloud-native applications. Developers can build and deploy applications that handle sensitive data from end to end in a confidential environment, enabling use cases in industries where data privacy is paramount.
SEV is completely hardware-based, so customers can enable Confidential Computing with a single click in their Google Cloud console.
Learn more about how Confidential Computing can mitigate data privacy concerns, while accelerating innovation.