Fortanix, a leading company in data-centric cybersecurity and confidential computing, has announced a strategic partnership with HPE and NVIDIA to bring its Armet AI platform to HPE Private Cloud AI solutions, co-developed with NVIDIA. The goal: to build secure, scalable “AI factories” that meet the most stringent regulations, whether on-premise or in hybrid and sovereign environments.
The integration enables deployment of Armet AI on HPE ProLiant Compute DL380a Gen12 servers equipped with NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, combining cutting-edge AI acceleration with advanced confidential computing capabilities to protect data, models, and prompts throughout the AI lifecycle.
Powerful AI… but with privacy and compliance by design
More and more companies want to harness generative and agentic AI capabilities, but they face three major barriers: privacy, data sovereignty, and regulatory compliance. Especially in regulated sectors (finance, healthcare, public sector, critical industries), training powerful models is no longer enough; demonstrating that data remains protected even during processing is essential.
This is where Fortanix Armet AI comes into play:
- It leverages NVIDIA Confidential Computing to run AI workloads within secure enclaves, where data, models, and prompts stay encrypted “in use”.
- It integrates with HPE Private Cloud AI, HPE’s turn-key “AI factory,” designed for governments and large enterprises to deploy high-performance AI under their own security and sovereignty conditions.
- It provides a unified layer of encryption, key management, and data governance aimed at simplifying compliance with frameworks like GDPR, NIS2, and emerging AI-specific regulations.
As Fortanix sums up, AI “only truly works” at scale when data, models, and prompts are secure by design, not as an afterthought.
Confidential AI factories: from the cloud to sovereign data centers
The concept of “AI factory” is consolidating as the new fundamental infrastructure unit: clusters of compute specifically for AI, with integrated storage, networking, and security, designed to train, tune, and serve models at scale.
Through this partnership, Fortanix, HPE, and NVIDIA offer a common foundation for these AI factories:
- End-to-end confidential compute
Armet AI protects data as it is stored, moved across the network, and most critically, while it is being processed on GPUs. This reduces the risk of leaks, industrial espionage, or unauthorized access to corporate models and sensitive prompts. - Infrastructure ready for generative and agentic AI
HPE ProLiant DL380a Gen12 servers with NVIDIA RTX PRO 6000 Blackwell GPUs supply the necessary power for state-of-the-art models, with a focus on energy efficiency and hardware-level security. - Flexible deployment: on-prem, hybrid, and sovereign “AI factories”
The joint architecture is designed to operate in private data centers, hybrid environments, or in new “sovereign AI factory” models, where data never leaves a specific jurisdiction and residency requirements are respected.
From demo to regulated environment: less risk, more traceability
Beyond technical safeguards, the joint approach also aims to streamline the day-to-day operations for security and compliance teams:
- Verifiable attestation: ability to demonstrate that AI workloads run on trusted hardware and isolated environments—crucial for auditors and regulators.
- Policy-based key and secret access management: centralized control over keys, secrets, and data access, with policy enforcement.
- Intellectual property protection: corporate models and weights can remain within confidential enclaves, reducing IP theft risks.
- Consistent security across the board: same confidential computing approach whether on-premise, public cloud, or in future “AI factories,” avoiding the complexity of multiple security models.
For organizations building internal AI platforms, the message is clear: it’s not just about deploying GPUs, but about creating a trustworthy AI platform where each generated token avoids adding new legal or security risks.
HPE, NVIDIA, and Fortanix: a trio for “serious” AI
This partnership is part of HPE’s Unleash AI program, aligning with the company’s move to position HPE Private Cloud AI as a leading option for governments and regulated industries seeking high-performance AI with full control.
NVIDIA contributes not only the Blackwell GPUs with NVIDIA Confidential Computing support but also their comprehensive AI software ecosystem, now enhanced with enclave execution capabilities.
Fortanix, meanwhile, offers the data security and confidential computing layer they’ve been advancing with leading tech giants, now bringing this directly into the realm of enterprise and sovereign AI factories.
In a landscape driven by generative AI growth, regulatory pressures, and debates on technological sovereignty, the combination of accelerated compute, hybrid cloud, and confidential computing aims to become a cornerstone of next-generation AI infrastructure.
via: fortanix