Vast Data Introduced Confidential AI Computing Service
The service creates hardware-isolated environments for training AI models on sensitive data, with a Q1 2027 ship date.
Updated on Sept. 22, 2026 in Artificial Intelligence
Live Poll
Do security concerns currently stop you from using advanced artificial intelligence tools at your work?
Vast Data has announced DataEnclave, a confidential computing service designed to allow organizations to run artificial intelligence models against sensitive data within protected environments. The technology is currently announced, with general availability scheduled for the first quarter of 2027.
Why it matters
The service targets organizations that currently avoid externally hosted AI services due to strict regulations or concerns over the distribution of proprietary model weights. It enables users to maintain control over their data and intellectual property while leveraging scalable infrastructure.
DataEnclave utilizes Nvidia confidential computing technology to encrypt guest memory, graphics processing unit (GPU) memory, and NVLink traffic. Before decryption, the system requires cryptographic attestation to verify that the underlying hardware, software, and governing policies are compliant.
The players
Vast Data
An enterprise storage and data infrastructure company focused on high-performance flash storage for AI and large-scale data processing.
Nvidia
A semiconductor company that designs high-performance graphics processing units and data center computing platforms for AI research and deployment.
The details
DataEnclave creates hardware-isolated environments that span both central processing units and graphics processing units (GPUs — specialized circuits designed to accelerate image processing and AI tasks). The system uses cryptographic attestation, a process where a secure module provides proof of identity and integrity to verify hardware and software before any model or data decryption occurs. All attestation events and enclave activity are logged within the Vast DataBase warehouse.
Timeline
September 22, 2026: Vast Data introduced the DataEnclave service.
First quarter of 2027: DataEnclave is scheduled to ship.
The Tech Race
DataEnclave follows the industry movement toward hardware-level security, aligning with Nvidia's broader roadmap to secure AI infrastructure against unauthorized access. This puts Vast Data in direct competition with other secure AI orchestration platforms aiming to solve the proprietary data bottleneck in enterprise machine learning.
The service will be included as part of the existing Vast Data AI Operating System at no additional cost for users. Enterprise organizations and developers will be able to access the environment starting in the first quarter of 2027.
The takeaway
DataEnclave aims to lower the barrier for regulated industries to adopt AI by automating the security of sensitive model weights. Watch for the official shipment in early 2027 to see how partner companies like Cohere and CrowdStrike integrate the platform into their production workflows.
Further reading
For broader trends in secure model deployment, see the latest developments in Artificial Intelligence.
Live Poll
Do security concerns currently stop you from using advanced artificial intelligence tools at your work?
