Google Cloud has unveiled a new security architecture for its serverless computing platform that uses confidential computing technology to encrypt data not only at rest and in transit but also during processing. The announcement, made at the Google Cloud Security Summit, represents a significant step toward eliminating one of the last remaining gaps in cloud data protection.
The Confidential Computing Paradigm
Traditional encryption protects data when it is stored on disk and when it moves across networks, but data must be decrypted before it can be processed. This creates a window of vulnerability during which sensitive information exists in plaintext in system memory, accessible to anyone with sufficient access to the host system, including the cloud provider itself.
Google Cloud’s new architecture, called Shielded Serverless, uses hardware-based trusted execution environments (TEEs) to process data within encrypted enclaves. Even Google engineers with administrative access to the underlying infrastructure cannot view or access data being processed within these enclaves.
Technical Implementation
Shielded Serverless leverages AMD’s Secure Encrypted Virtualization with Secure Nested Paging (SEV-SNP) technology. Each serverless function execution receives its own encrypted memory space, isolated from all other workloads and from the host operating system. Cryptographic attestation allows customers to verify that their code is running in a genuine, unmodified enclave before sending sensitive data for processing.
Use Cases and Early Adopters
The technology is particularly relevant for industries that handle sensitive data under strict regulatory requirements. Healthcare organizations can process patient data in compliance with HIPAA without trusting the cloud provider with plaintext access. Financial institutions can run risk models on sensitive trading data without exposure. Government agencies can use cloud computing for classified workloads that were previously restricted to on-premises infrastructure.
“Confidential computing removes the last major objection to cloud adoption for our most sensitive workloads,” said the CISO of a major European bank that participated in the beta program. “We no longer have to trust the cloud provider not to look at our data. The hardware enforces the guarantee mathematically.”
Competitive Landscape
Google is not the first cloud provider to offer confidential computing capabilities. Microsoft Azure introduced confidential VMs in 2022, and AWS launched its Nitro Enclaves platform even earlier. However, Google’s implementation is the first to extend confidential computing to serverless functions, which are increasingly popular for event-driven and microservices architectures.
Performance Considerations
Early benchmarks show that Shielded Serverless functions experience a 3 to 8 percent performance overhead compared to standard serverless execution, primarily due to the encryption and decryption operations required to move data into and out of the enclave. Google engineers said they expect this overhead to decrease as hardware support for confidential computing continues to improve.
Shielded Serverless is available immediately in preview for Google Cloud customers in the United States and Europe, with general availability planned for Q4 2026. Pricing includes a modest premium over standard Cloud Functions, reflecting the additional hardware resources required for confidential execution.
