The sovereignty imperative: Why AI leaders are taking control of their intelligence

As AI becomes a core business capability, a new strategic imperative has emerged for CIOs: sovereign AI. The next phase of competitive advantage will belong to organizations that control their intelligence stack, including where data resides, how models are governed, and who owns the intellectual property they generate.

The urgency is growing. In a recent global survey, 71% of executives, investors, and government officials described sovereign AI as either an existential concern or a strategic imperative for their organizations.

The risks of outsourced intelligence are no longer theoretical. Evolving regulations, geopolitical uncertainty, and limited visibility into third-party AI systems are making sovereign infrastructure a prerequisite for mission-critical workloads.

The CIO’s sovereignty mandate

Sovereign AI is not simply about keeping data within a country’s borders. It is a layered strategy spanning technology, policy, and business autonomy. For a CIO, it means demanding accountability across distinct dimensions:

• Infrastructure sovereignty: Running systems on-premises or within controlled environments to avoid reliance on foreign-hosted platforms.
• Data sovereignty: Ensuring data is processed in compliance with local laws like GDPR or HIPAA, protecting proprietary IP from leaking into public training sets.
• Model sovereignty: Retaining full control over model weights and architecture to ensure AI reflects specific organizational context and cultural values.
• Operational autonomy: The ability to operate AI systems independently of external APIs, ensuring business continuity during vendor or geopolitical disruptions
• Financial autonomy: The ability of an organization, institution, or nation to maintain control over its financial assets, transactions, data, and decision-making without undue dependence on external entities.

Sovereign AI is already helping organizations — from accelerating scientific discovery and advancing healthcare research to protecting national interests and safeguarding valuable intellectual property. The following examples illustrate how organizations are leveraging AI infrastructure co-engineered by HPE and NVIDIA to maintain control over their data, models, and operations.

Case Study 1: The University of Utah and the State of Utah

The challenge: Accelerating medical research, AI innovation, and regional economic growth while maintaining strong governance over sensitive healthcare and research data. The University of Utah and the State of Utah have invested $50 million in a new AI factory designed to expand access to advanced AI infrastructure for researchers, industry partners, and public-sector organizations. 

The initiative is expected to more than triple the university’s computing capacity, enabling increasingly data-intensive research and AI workloads. But the investment is about more than scale. It provides a secure environment where researchers can work with complex biomedical datasets, develop new AI models, and collaborate across academia, government, and industry while maintaining oversight of how data is accessed, managed, and protected. 

The result is a platform that supports healthcare and life sciences research, AI innovation, startup collaboration, and workforce development. By bringing advanced AI capabilities closer to the researchers, institutions, and businesses that depend on them, the University of Utah is demonstrating how AI infrastructure can support both scientific advancement and regional economic growth.
 
Case Study 2: Argonne National Laboratory

The challenge: Advancing scientific discovery and national research priorities while maintaining governance over critical research data, models, and intellectual property. At Argonne National Laboratory, AI infrastructure is a strategic asset for scientific leadership. 

The laboratory’s Janus and Tara systems combine HPE Cray Supercomputing technology with NVIDIA accelerated computing to support large-scale AI training, inference, and simulation workloads. For organizations operating at the forefront of research, the challenge is not simply accessing more compute. It is ensuring that critical research data, scientific workloads, and resulting discoveries remain under appropriate governance. 

Deployed within a federal research environment, the Janus and Tara systems help researchers pursue breakthroughs across disciplines such as climate science, materials discovery, and advanced manufacturing while maintaining stewardship of the underlying data, models, and research outcomes. The deployment highlights an increasingly important reality for CIOs and technology leaders: as AI becomes central to innovation, control over where data resides, where models are trained, and how intellectual property is protected can be just as important as computational performance.
 
Case Study 3: Bosch and the Race to Autonomous Driving

The challenge: Accelerating AI-driven product development while managing large volumes of proprietary engineering and simulation data across global operations. At Bosch, AI has become a key enabler of autonomous-driving innovation. The company uses large-scale simulation and virtual testing environments to develop, validate, and refine advanced driving systems before they reach public roads. These efforts depend on massive volumes of sensor, simulation, and vehicle data, making governance, intellectual property protection, and operational control critical to innovation at scale.

To support this effort, Bosch reengineered its AI software development environment to scale collaboration and support fleets of virtual vehicles that can be tested thousands of hours each day in simulation. Engineers can continuously evaluate new driving scenarios, accelerating development and validation efforts before vehicles ever reach public roads. Governance and control play a critical role in that strategy. 

By maintaining oversight of the environments used to develop and validate these systems, Bosch can better protect valuable intellectual property, support collaboration across global engineering teams, and address regional data governance requirements without sacrificing innovation speed. For global manufacturers, this illustrates an increasingly common reality: AI success depends not only on scaling compute resources, but also on maintaining control over the data, models, and proprietary knowledge that drive competitive advantage.

Industrializing the AI lifecycle: The HPE AI Factory lab

To help CIOs validate these strategies, HPE and NVIDIA established an AI Factory Lab in Grenoble, France. This is where organizations can test and refine workloads on a sovereign infrastructure stack.

The lab is equipped with engineering-validated architecture that includes the latest NVIDIA AI Enterprise government-ready software, HPE servers, NVIDIA Spectrum-X Ethernet networking, and HPE storage. By hosting this infrastructure within the EU, the facility directly addresses the needs of global enterprises for data sovereignty and regional regulatory compliance. It allows customers to benchmark their workloads, estimate power and cooling requirements, and understand the total cost of ownership before committing capital to a full-scale deployment.

The technical platform for sovereignty

Creating a sovereign AI factory requires more than just high-performance hardware. It requires a composable and validated solution that can handle heterogeneous workloads throughout the entire AI lifecycle—from data ingestion and model training to high-volume inference and monitoring.

Key technical requirements for these factories include:

1. Hard multi-tenancy: Physically segregated compute nodes and network isolation to ensure that multiple user groups or departments can share infrastructure without data leakage.
2. Air-gapped management: For highly regulated sectors, the ability to operate in network-isolated environments is critical for maintaining absolute data privacy.
3. Lifecycle operations: Integrated tools like HPE Morpheus and HPE OpsRamp provide deep observability and control, allowing IT teams to manage the deployment, maintenance, and retirement of components efficiently.
4. Security-first design: Incorporating features like STIG-hardened and FIPS-enabled software, as well as post-quantum cryptography, to protect the entire stack from emerging threats.

Own your intelligence

For today’s CIO, the message is clear: AI strategy is increasingly becoming infrastructure strategy. Whether the goal is accelerating scientific discovery, advancing healthcare research, protecting intellectual property, or supporting global innovation, organizations need the ability to govern how AI is built, deployed, and operated. 

Together, HPE and NVIDIA are helping organizations establish the foundation for sovereign AI through integrated infrastructure, accelerated computing, enterprise software, and operational expertise. By combining AI factory solutions with the governance, security, and control required for mission-critical workloads, they are enabling organizations to scale AI without surrendering ownership of their data, models, or intellectual property. 

The next era of competitive advantage will not belong to organizations that simply adopt AI. It will belong to those that own, govern, and scale it on their terms. The question for CIOs is no longer whether AI will become a strategic asset. The question is whether the intelligence powering the business will ultimately belong to you.

3 actions CIOs can take now 

As sovereign AI moves from strategic discussion to operational reality, CIOs can focus on three priorities.

1. Treat AI infrastructure as a strategic asset. Many organizations still approach AI as a collection of individual projects. Instead, evaluate AI as a long-term capability that requires governance, security, power, cooling, networking, and lifecycle management. The organizations creating sustainable competitive advantage are building AI factories designed to support a growing portfolio of applications, models, and agents.

• Assess your sovereignty posture across the entire AI stack. Data sovereignty is only one piece of the equation. CIOs should evaluate where data resides, who owns model weights, which third parties have access to critical workloads, and how operations can be affected by regulatory, geopolitical, or vendor-driven disruptions. An AI strategy is only as resilient as the infrastructure and governance frameworks that support it.

• Establish a roadmap for sovereign AI at scale. Sovereignty should not be viewed as a future requirement reserved for governments or highly regulated industries. Organizations should begin defining policies for data governance, model ownership, infrastructure control, and operational autonomy today. Starting with a clear roadmap enables the business to scale AI confidently while maintaining control over the intelligence it creates.

Engage HPE Services for a sovereign AI workshop and learn more about HPE Sovereign AI Factory at hpe.com/ai/sovereign.

*************

As AI becomes increasingly central to economic competitiveness, scientific advancement, and national priorities, organizations require infrastructure that balances performance with security and sovereign control. Together, HPE and NVIDIA co-engineer rack-scale AI systems that integrate AI computing, high-performance networking, and supercomputing expertise to support large-scale AI workloads. This provides enterprises, governments, and research institutions with a trusted foundation for sovereign AI initiatives while maintaining control over critical data, models, and operations.