Vast Data has unveiled DataEnclave, a new confidential computing environment integrated into its AI Operating System. The platform utilizes Nvidia technology to establish hardware-isolated spaces, allowing organizations to process sensitive data and proprietary AI models without exposing them to infrastructure operators or third parties.
The system secures memory across CPUs and GPUs, encrypting NVLink traffic and guest memory. It employs a cryptographic attestation process to verify hardware and software configurations before decryption. Users maintain control over their encryption keys through their own management systems, with all lifecycle activities recorded in the Vast DataBase.
DataEnclave supports deployment in private data centers, AI clouds, and air-gapped environments. It leverages the open-source Cloud Native Computing Foundation Trustee project or Fortanix Confidential AI for attestation. The service is designed to allow base models and fine-tuned weights to interoperate while remaining under separate ownership.
Initial partners include Cohere, CrowdStrike, Deepgram, The San Francisco AI Factory, Fundamental Research Labs, and TwelveLabs. Hardware integration is provided by Cisco Systems and Super Micro Computer. The company is targeting highly regulated sectors, including defense, intelligence, banking, and pharmaceuticals.
The new capability is included within the existing Vast AI Operating System at no additional cost. DataEnclave is currently available in preview, with a full release scheduled for the first quarter of 2027 through Vast Data and its hardware partners.
