Confidential Computing

Definition

Confidential computing is a security technique that protects data while it is actively being processed in memory by running computations inside a hardware-based Trusted Execution Environment (TEE), keeping data encrypted and inaccessible even to the cloud provider, hypervisor, or operating system hosting it.

Overview

Traditional encryption protects data at rest (on disk) and in transit (over the network), but data has historically had to be decrypted in memory to be processed by the CPU — creating a window where a compromised hypervisor, malicious cloud administrator, or kernel-level attacker could potentially access sensitive plaintext data or the encryption keys protecting it. Confidential computing closes this gap by using CPU-level hardware features to create a Trusted Execution Environment (TEE): an isolated, encrypted memory region where code and data remain encrypted even while being actively processed, with the CPU itself managing decryption only within its own protected boundary. Major hardware implementations include Intel SGX (Software Guard Extensions, enclave-based) and its successor Intel TDX (Trust Domain Extensions, VM-based), AMD SEV-SNP (Secure Encrypted Virtualization), and ARM's Confidential Compute Architecture — each isolating a workload's memory from the hypervisor, OS, and even physical access to RAM (mitigating cold-boot and DMA attacks). Cloud providers expose these capabilities as confidential VM offerings (Azure Confidential Computing, Google Confidential VMs, AWS Nitro Enclaves), and the technique typically includes remote attestation: a cryptographic proof that a workload is running unmodified inside a genuine, correctly configured TEE, which a remote party can verify before releasing sensitive data or keys to it. The Confidential Computing Consortium, a Linux Foundation project with members including Intel, Microsoft, Google, AMD, Meta, and Red Hat, works to standardize the space and promote adoption. Practical use cases center on scenarios requiring processing sensitive data in environments the data owner doesn't fully trust: multi-party data analysis where competing organizations want to run joint computation without exposing raw data to each other, processing regulated data (healthcare, financial) in public cloud environments with stronger isolation guarantees, protecting AI model weights and inference data from the infrastructure provider, and blockchain/Web3 applications needing verifiable off-chain computation. Confidential computing complements, rather than replaces, encryption at rest and in transit — together the three form a more complete 'data always encrypted' posture across its full lifecycle.

Key Concepts

• Protects data in use, closing the gap left by encryption at rest and in transit
• Uses hardware Trusted Execution Environments (TEEs) with CPU-managed encrypted memory
• Major implementations: Intel SGX/TDX, AMD SEV-SNP, ARM Confidential Compute Architecture
• Isolates workloads from the hypervisor, OS, and cloud provider administrators
• Remote attestation cryptographically proves a workload is running unmodified inside a genuine TEE
• Available as managed cloud offerings (Azure Confidential Computing, AWS Nitro Enclaves, Google Confidential VMs)
• Standardized and promoted by the Linux Foundation's Confidential Computing Consortium
• Enables multi-party computation without exposing raw data between parties

Use Cases

Frequently Asked Questions

Related in Cybersecurity