Data has always been vulnerable at three points: at rest, in transit, and in use. The first two have well-established defenses-encryption at rest and TLS for data moving across networks are standard practice. The third, protecting data while it's actively being processed, has remained the hardest problem in enterprise security. IBM's expansion of Hyper Protect capabilities onto Red Hat OpenShift and Red Hat Virtualization platforms, running on IBM Z and LinuxONE hardware, targets precisely that gap.
The new offerings-Hyper Protect Confidential Containers for OpenShift and Hyper Protect Container Runtime for Virtualization-extend what's known as confidential computing into environments that many large organizations already run their infrastructure on. The core idea is straightforward even if the engineering behind it is not: workloads, including the memory they use while executing, are isolated and encrypted in a way that shields them from the underlying infrastructure, the hypervisor, and even administrators with privileged access. This matters enormously for multi-tenant cloud setups and shared infrastructure, where questions about data isolation are a constant concern-much like how shared IP addresses on networks raise distinct trust problems, including why shared IPs get flagged by security systems that struggle to distinguish between legitimate and malicious traffic sharing the same resource.
Why "In Use" Protection Changes the Security Calculus
Traditional security models assume that whoever controls the hardware and hypervisor layer is trustworthy by default. Confidential computing rejects that assumption. By relying on IBM Secure Execution for Linux, these new capabilities create hardware-enforced boundaries around workloads, meaning that even a compromised or malicious administrator cannot inspect or tamper with the data being processed inside a protected container. For industries bound by strict regulatory regimes-financial services, healthcare, government contracting-this shifts the compliance conversation. Instead of relying purely on policy and audit trails, organizations gain a technical guarantee that sensitive processes, including proprietary AI models and the data that trains them, remain sealed off from unauthorized access.
Built for a Zero-Trust, AI-Heavy Landscape
The timing is not incidental. As organizations deploy more AI models in production and increasingly operate across hybrid and multi-cloud environments, the attack surface for intellectual property theft and data leakage has grown substantially. Zero-trust architecture-where no component of the system is automatically trusted-has become the operating assumption for security teams rather than an aspiration. Embedding Hyper Protect directly into Red Hat's ecosystem means enterprises already standardized on OpenShift or Red Hat Virtualization do not need to rebuild their infrastructure from scratch to gain this layer of protection. It integrates with workflows they already use.
What This Means Going Forward
Confidential computing is still maturing as a category, and adoption will depend on how well it balances security guarantees against performance overhead and operational complexity. IBM Z and LinuxONE have long positioned themselves around resiliency and security for mission-critical workloads, so this move extends an existing strength rather than representing a departure. For enterprises weighing how to protect AI models, customer data, and regulated workloads simultaneously, the ability to isolate processing at the hardware level-without abandoning familiar platforms-offers a pragmatic route toward meeting both security and compliance demands that are only going to intensify.