Fortanix Inc. has integrated its Confidential AI platform with the Cloud Native Computing Foundation’s Confidential Containers project. This update allows enterprises to execute artificial intelligence workloads on Kubernetes within hardware-based trusted execution environments.
The solution addresses security risks during AI inference, when data and model weights are decrypted in processor memory. Fortanix reported that 93% of 300 surveyed AI leaders at large U.S. companies believe traditional encryption fails to protect data during processing, with 41% having blocked deployments due to these concerns.
The new plugins verify the authenticity of containers before granting access to encryption keys. This process is managed by the Fortanix Confidential Computing Manager, which performs joint attestation for CPUs and GPUs. Keys are released only if security policies are met, ensuring they remain within protected environments.
The platform supports on-premises and air-gapped deployments, catering to regulated industries with data sovereignty requirements. It enables companies to host proprietary models alongside their own data without compromising intellectual property. The system is designed to remain operational even if an individual server fails.
Fortanix lists over a dozen supporting companies, including Advanced Micro Devices, Intel, CrowdStrike, and Nvidia. The platform is available now, and the company plans to demonstrate it on Nvidia GPUs at the Nvidia GTC Berlin event, scheduled for October 20–22.
