VAST Data Launches DataEnclave for Confidential Enterprise AI
The AMW Read
DataEnclave meaningfully expands VAST's infrastructure role from data systems into confidential model execution for regulated enterprise deployments.
VAST Data Launches DataEnclave for Confidential Enterprise AI
VAST Data has introduced DataEnclave, a confidential AI system for running models on sensitive corporate data without exposing customer data or model weights to infrastructure operators. The product extends VAST's AI Operating System and DataEngine with a hardware-isolated runtime, cryptographic attestation, separate enterprise and model-provider key management, and audit logging. It is built on NVIDIA confidential-computing capabilities across Hopper, Blackwell, and Rubin platforms, protecting CPU and GPU memory as well as NVLink traffic.
The launch addresses a central adoption constraint for regulated enterprises: data owners may not be able to send sensitive information to external AI services, while model owners may be unwilling to deploy weights into environments they do not trust. By making the runtime attestable before decryption keys are released, VAST is positioning its infrastructure stack as a control plane for both data and model IP. The move also extends VAST's September partnership with CrowdStrike around securing AI data pipelines, from telemetry and access security toward protected model execution.
For builders, the practical test is whether attestation, key separation, and operational auditability can be integrated without making agentic or model-serving workflows too difficult to operate. For investors, the product highlights how infrastructure vendors can compete beyond storage performance by owning the trust boundary for sovereign, air-gapped, and compliance-sensitive AI deployments. VAST Data is tracked as an AI Infrastructure company with about $620M in total funding, per the AI Market Watch index, which covers roughly 5,000 companies rather than a census.
