Enclaive has unveiled a comprehensive approach to securing cloud and AI services by integrating confidential computing with verifiable VM firmware, key management, and protected AI workflows. The German company argues that simply deploying an isolated virtual machine leaves critical gaps in the trust chain, particularly around firmware integrity, cryptographic key handling, and the processing of sensitive data by AI models.
The newly introduced Sylica firmware is purpose-built for confidential computing environments. It includes only the functions required by virtual machines, thereby reducing the attack surface. Reproducible builds allow independent researchers and auditors to examine and attest the firmware, enabling this layer to be incorporated into the overall trust verification process rather than accepted on faith.
CEO Andreas Walbrodt described Sylica as a market-first solution that completes confidential computing from the confidential VM firmware level upward. The technology will be showcased at the it-sa trade fair.
Complementing the firmware is the Enclaive Multicloud Platform (eMCP), which automates the provisioning and management of confidential computing workloads. It supports both VM-based and Kubernetes-based deployments and connects multiple cloud providers with different hardware security technologies, addressing the operational complexity of running protected workloads across heterogeneous infrastructures.
Key management is handled by a virtual HSM (vHSM) that acts as a cloud-native cryptographic engine. It supports external key management, independent attestation of protected environments, and post-quantum encryption levels.
For AI-specific use cases, the Garnet platform pseudonymizes sensitive information before it is passed to external AI models. It also enables model execution inside isolated environments and secures confidential RAG workflows.
