VAST Data Launches DataEnclave to Run Enterprise AI Anywhere with Confidential Computing

VAST Data has introduced DataEnclave, a confidential computing service designed to resolve a fundamental challenge facing enterprise AI: how to run proprietary AI models on sensitive data that cannot leave tightly controlled environments1

1

2

2

. The new capability within the VAST AI Operating System uses NVIDIA Confidential Computing to create hardware-isolated environments where both enterprise data and model weights remain protected during processing3

3

.

Addressing the Enterprise AI Deadlock

Across financial services, healthcare, government agencies, and other highly regulated industries, valuable data remains locked inside environments where data sovereignty rules, regulatory constraints, and security requirements prevent it from moving to external AI services1

1

. Phil Manez, VP of strategic initiatives at VAST Data, explains that "organizations like banks, government agencies, and health care providers have a lot of data that was never intended to move to the cloud"1

1

. Meanwhile, model builders have been reluctant to distribute their proprietary models into infrastructure they don't control, creating an impasse that has kept the most advanced AI models away from the most sensitive data environments2

2

.

How Confidential AI Protects Both Sides

DataEnclave addresses both concerns simultaneously through cryptographic attestation and trusted execution environments spanning CPUs and GPUs2

2

. The system encrypts guest memory, GPU memory, and NVLink traffic while isolating active models and data from administrators, infrastructure operators, and other tenants2

2

. Before any sensitive assets are decrypted, the attestation process verifies the hardware, software, and governing policies3

3

. Customer data keys remain under customer control, while model keys and weights stay within the model builder's trust domain3

3

.

Jeff Denworth, VAST Data's co-founder, emphasizes the stakes: "Model weights are fast becoming the most valuable intellectual property in the world. Base weights define the value of frontier models, while fine-tuned weights will increasingly represent the proprietary intelligence of AI-driven enterprises"2

2

.

Protecting Data Security During Processing

While encrypting data at rest and in transit has been possible for years, securing it in memory during processing has remained the biggest challenge1

1

. Confidential computing isolates working memory from the host operating system, hypervisor, and administrators with root access1

1

. For AI workloads, this protection extends to GPUs, where training and inference put model weights, prompts, and intermediate results into memory while they're in use1

1

.

The model inside the enclave never accesses proprietary records directly. Instead, a retrieval application outside the enclave searches the enterprise's documents, assembles relevant excerpts into a prompt, and sends that prompt to the model's inference API1

1

. The provider that owns the model gets no administrative path into the enclave to read that prompt or the response1

1

.

Flexible Deployment and Ecosystem Support

DataEnclave can be deployed in customer data centers, AI clouds, and fully air-gapped environments2

2

. Initial model and software partners include Cohere, CrowdStrike, Deepgram, The San Francisco AI Factory, Fundamental Research Labs, and TwelveLabs2

2

. Hardware partners Cisco Systems and Super Micro Computer will offer integrated systems2

2

.

James Manning, CEO of Sharon AI, a trusted AI infrastructure provider, explains the value for sovereign operations: "Building on our sovereign data foundation with VAST, DataEnclave lets us host those models onshore, inside attested environments where the model owner's weights and the customer's data are both protected from everyone, including us"3

3

.

Changing Enterprise AI Economics

VAST Data argues that DataEnclave could transform the economics of running enterprise AI anywhere by allowing customers to run proprietary AI models on infrastructure they own or select a regional AI cloud, reducing model developers' need to finance underlying computing capacity2

2

. The capability will be included in the existing AI Operating System at no additional charge2

2

.

DataEnclave is available in preview and scheduled to ship in the first quarter of 2027 through VAST Data and participating hardware partners2

2

. The system logs attestation events, key releases, and enclave lifecycle activity in the VAST DataBase data warehouse, providing an audit trail that shows which workloads ran and under what policies without revealing protected data or model weights2

2

.