Azure Confidential Computing for Sensitive Aged Care and Financial Workloads

Most organisations that handle regulated personal, health, and financial data have invested heavily in encryption at rest and encryption in transit. These controls protect information while it sits in storage and while it moves across networks, and they form a reasonable baseline for any modern data protection programme. They do not, however, address the moment when data is actually being processed, because information must be decrypted into memory before a workload can operate on it.

This gap matters for aged care providers processing clinical records and for financial services organisations handling account, identity, and transaction data. When sensitive information sits in plaintext in system memory, it becomes exposed to a range of threats, including compromised hypervisors, privileged administrators, and malicious insiders with infrastructure access. Azure Confidential Computing is designed to close this gap by protecting data while it is in use, and understanding how it works helps technical and decision-making readers evaluate whether it belongs in their architecture.

What data in use protection actually means

Confidential computing protects data during processing by isolating it inside a hardware-based trusted execution environment, often shortened to a TEE. The processor encrypts the memory associated with a protected workload and enforces boundaries that prevent code outside the environment from reading it, including the host operating system and the cloud platform itself. This means the confidentiality of the workload no longer depends solely on trusting the underlying infrastructure operator or the hypervisor stack.

The practical consequence is a materially reduced attack surface for the most sensitive stages of processing. An administrator with elevated privileges on the host, or an attacker who compromises the virtualisation layer, cannot simply read the contents of protected memory. For regulated workloads, this supports a stronger separation of duties and gives organisations a defensible technical control to place around data that legislation and sector standards treat as high risk.

Confidential virtual machines and confidential containers

Azure offers confidential computing through two primary models. Confidential virtual machines encrypt the entire memory space of the VM using hardware technologies such as AMD SEV-SNP and Intel TDX, allowing organisations to lift and shift existing workloads with relatively little redesign. This model suits established line of business applications, database workloads, and clinical or financial systems that cannot easily be re-architected but still require memory-level isolation.

Confidential containers extend the same principle to containerised workloads, including those running on Azure Kubernetes Service and Azure Container Instances. This approach appeals to teams that have already adopted a microservices or container strategy and want to apply confidentiality selectively to the services that process regulated data. Choosing between the two models generally depends on how an application is packaged today, how much refactoring is acceptable, and which components genuinely need memory-level protection rather than blanket application across an entire estate.

Attestation and establishing trust

Encrypting memory is only part of the picture, because an organisation also needs assurance that a workload is genuinely running inside a legitimate trusted execution environment before it releases sensitive data or secrets to that workload. This assurance comes from remote attestation, a process in which the hardware produces a cryptographically signed report about the state of the environment. A verifier such as Microsoft Azure Attestation examines that evidence and confirms the environment is authentic and configured as expected.

For aged care and financial services architects, attestation is what turns confidential computing from a marketing concept into an enforceable control. It allows a system to make release of keys or data conditional on a verified environment, so decryption keys are only made available once the platform confirms the workload is running in a genuine, unaltered TEE. Designing attestation into key management and secrets handling early is far easier than retrofitting it, and it produces evidence that can support audit and assurance activities.

Regulatory context for Australian organisations

Australian aged care providers and financial services organisations operate under overlapping obligations that reward strong technical controls over sensitive data. The Privacy Act and the Australian Privacy Principles require organisations to take reasonable steps to protect personal information, and health information carries additional sensitivity under that framework. Financial institutions are also subject to prudential expectations, including the information security requirements set out in APRA Prudential Standard CPS 234, which emphasises protecting information assets in line with their sensitivity and criticality.

Confidential computing does not, on its own, satisfy any of these obligations, and it should be understood as one control within a broader security and governance posture. It can, however, strengthen the case that an organisation has taken meaningful steps to protect regulated data throughout its lifecycle, including during processing. Guidance from the Australian Cyber Security Centre, particularly the Essential Eight and cloud security material, remains the appropriate reference point for building the surrounding controls that confidential workloads still depend upon.

Practical considerations before adoption

Adopting confidential computing works best when it is targeted rather than universal. Organisations should begin by identifying the specific workloads and data flows where memory-level exposure represents a genuine risk, such as processing of health assessments, identity verification, or transaction enrichment, rather than attempting to protect everything at once. This focus keeps costs proportionate and avoids the operational overhead of running confidential infrastructure where it delivers little additional assurance.

Teams should also account for practical factors including regional availability of the required hardware in Australian datacentres, compatibility of guest operating systems and container runtimes, and the effect on existing deployment pipelines and monitoring. Testing performance under realistic load is sensible, as is validating that attestation, key management, and logging integrate cleanly with existing platforms such as Azure Key Vault and Microsoft Entra. Approached deliberately, confidential computing gives regulated organisations a credible way to protect their most sensitive data at the one point where traditional encryption leaves it exposed.

References

• Azure Confidential Computing overview (Microsoft Learn)
• Confidential virtual machines overview (Microsoft Learn)
• Essential Eight (Australian Cyber Security Centre)