Fortanix Adds Confidential Containers Support for Kubernetes AI Workloads

Fortanix announced that its Confidential AI platform supports CNCF Confidential Containers (CoCo), enabling organizations to run AI workloads on Kubernetes inside hardware-based Trusted Execution Environments (TEEs), using confidential CPUs and GPUs. The company says the approach preserves the speed and agility of containers while protecting data, models, agents, and applications while they are in use. The capability is available now and is positioned for organizations seeking to scale AI while protecting sensitive information and intellectual property, including customers in government and other regulated sectors.

The security model centers on isolating workloads in hardware-based TEEs and using attestation to verify an environment before data or models are unlocked. Fortanix describes controlled key release as a way to ensure that access depends on verification of the trusted environment. The company also says the platform can support proprietary models deployed on-premises or in air-gapped environments, close to sensitive data, without exposing model owners’ intellectual property to the infrastructure operator. These are product claims in a commercial announcement; the source does not identify government certifications, specific agency requirements, or validated agency deployments.

For federal procurement teams and contractors, the announcement is a technology signal rather than an open buying opportunity. Organizations assessing Kubernetes-based AI workloads can examine whether CoCo’s open-standard approach, workload attestation, and key-release controls align with their security architecture and data-protection needs. The release does not provide a contract value, solicitation, procurement timeline, or deadline. It also does not establish that the product meets any particular federal authorization or compliance regime, so buyers would need to verify applicable certifications and requirements independently before considering deployment.

Fortanix framed the product against a broader challenge in AI security, citing research it conducted with Atomik Research: 93% of AI leaders surveyed understand that traditional encryption does not protect data while AI processes it, while only half said they can fully verify that protection is in place during inference. Fortanix co-founder and CEO Anand Kashyap said containers do not change the underlying threat because models, agents, and applications still interact with sensitive data and intellectual property. The release says teams can deploy containers into confidential-computing environments with one click across supported hardware and Kubernetes platforms, with workloads verified before keys are released. Fortanix plans to demonstrate the solution at NVIDIA GTC Berlin from October 20–22, 2026.

• Product availability: Fortanix says Confidential AI with CNCF Confidential Containers support is available now.
• Deployment context: The capability targets Kubernetes AI workloads running in hardware-based TEEs on confidential CPUs and GPUs.
• Security features: The company highlights environment attestation and controlled key release before data or models are unlocked.
• Sensitive-data use cases: The announcement identifies government and other regulated sectors, as well as on-premises and air-gapped deployments, as relevant contexts.
• Research figures: Fortanix and Atomik Research report that 93% of surveyed AI leaders recognize the limits of traditional encryption during processing, while only half can fully verify protection during inference.
• Procurement posture: The release describes a commercial product, not a government award, solicitation, or contract opportunity; no contract value or procurement deadline is provided.
• Due diligence: Federal buyers and contractors should independently confirm applicable certifications, agency requirements, supported hardware, Kubernetes compatibility, and operational controls.
• Demonstration: Fortanix plans to showcase the solution at NVIDIA GTC Berlin, October 20–22, 2026.
• Industry context: John Fanelli, NVIDIA’s Vice President of Enterprise Software, said, “Through open source collaboration with the CNCF Confidential Containers community, Fortanix is building on open standards for attestation and secure key release at production scale.”