Data security company Fortanix Inc. today announced a new joint solution with Nvidia Corp. that offers a turnkey, on-premises platform for running secure and sovereign, agentic artificial intelligence in AI Factories and highly regulated environments.

The new solution allows enterprises to run AI on sensitive data with verifiable trust and compliance that includes a confidential AI pipeline, composite attestation and hardware security module-gated key release.

The solution seeks to address the issue whereby industries such as healthcare, finance and government want to take advantage of the promise of AI but face a significant roadblock: The data they work with is too sensitive to move into the cloud without serious risks. Not just morally but legally, private identifiable data, healthcare records, financial transactions and classified information need to stay protected at all times.

Fortanix argues that before now, enterprises were often forced to choose between speed and innovation on one hand and compliance and control on the other.

The new solution, powered by Nvidia Confidential Computing, eliminates the tradeoffs by allowing organizations to go from AI pilots to production in days without compromising security, compliance or data sovereignty.

The solution runs on Nvidia Hopper and Nvidia Blackwell graphics processing unit architectures to give enterprises a trusted way to deploy AI agents on sensitive data. The platform enables organizations to unlock AI’s full potential without sacrificing trust or performance where compliance and confidentiality are nonnegotiable.

“AI factories are engines of enterprise transformation and must be built on a foundation of security and trust,” said Justin Boitano, vice president of enterprise AI products at Nvidia. “Together with Fortanix, we’re bringing Nvidia Confidential Computing into these next-generation environments, enabling organizations to run powerful AI with verified confidentiality, compliance and control in on-premises data centers.”

Key benefits of the Fortanix platform include end-to-end trust, compliance by design and faster time to production.

On the security side, the end-to-end trust sees Fortanix protecting data and models throughout the lifecycle with Nvidia Confidential Computing and composite attestation across central processing units and GPUs, leveraging Nvidia NRAS. The Fortanix Data Security Manager only releases cryptographic keys once workloads are verified to ensure that data, models and secrets remain protected.

Compliance by design means the platform, which is supported by a Federal Information Processing Standard 140-2 Level 3 Hardware Security Module, enforce strict key custody, role-based access controls and consistent audit logs. For quicker production, the Fortanix Armet AI allows enterprises to quickly deploy a turnkey AI platform on-premises, with built-in AI guardrails, observability and orchestration.

Fortanix is demoing the new platform at the Nvidia GTC conference in Washington, DC, through Oct. 29.

Photo: Fortanix