Sovereign AI in action

The recent Open Source Summit Hyderabad offered a platform to discuss a significant shift in the global AI landscape: the move toward a distributed, trusted, and self-reliant AI model. For too long, the narrative around AI has been dominated by a few centralized players and proprietary platforms. That era is quickly ending. India is now part of this movement with its unique, development-centered approach to AI. I had the pleasure of presenting with Rajgopal A S, MD and CEO at NxtGen Cloud Technologies. Our joint presentation, "Sovereign AI in Action," showcased how this vision is being realized through the powerful combination of NxtGen's SpeedCloud and Red Hat's open source technologies.

The imperative for sovereign AI in India

Organizations and nations increasingly recognize the critical need for digital sovereignty—the ability to independently control and safeguard essential digital infrastructure, data, operations, technology, and supply chains in alignment with their policies, values, and strategic objectives. This is particularly urgent in India due to several interconnected challenges:

• Data control and protection: Governments see sovereign AI as vital for protecting data privacy and safeguarding national security. This requires controlling where data is stored, who can access it, and how it’s used, often insisting on data staying within specific national or regional borders. The growth of data sovereignty laws and data protection rules worldwide means companies are having to adjust quickly.
• Technological independence: There's also a strong desire to reduce reliance on external providers' infrastructure and technologies amid geopolitical uncertainties. Currently, 95% of accelerator-enabled public cloud regions use US-owned accelerators, showing a concentration of control. Technology sovereignty covers data, hardware, software, and services, strengthening trust and enhancing security posture. According to this study, only 24 countries in the world today have access to accelerators powerful enough to train frontier AI models, severely limiting which countries can perform this with their own data.
• Operational autonomy: Maintaining control over standards, processes, and policies is vital for transparency and easy verification. It also enables systems to operate in disconnected or air-gapped environments, so vital services are more likely to remain active despite internet or supply chain issues.
• Cultural and localization relevance: Many AI models need to be specific to local languages and cultural contexts. By developing and training models on local data, AI systems are better able to reflect local societal norms and values, avoiding or countering potential biases from foreign models. In fact, the internet is actually mostly western content, which means specific cultural content like that in India is not available to be used in training state-of-the-art models, necessitating local efforts to plug this gap.
• India's vision for AI: Rajgopal A S, MD and CEO at NxtGen Cloud Technologies, described the India AI mission as, "pretty comprehensive in nature and no other country I'm aware of has got such a comprehensive approach to developing AI in the country." This public-sector led, development-focused initiative, supported by a $1.2 billion fund (10,500 crores) for GPU access, aims to democratize AI through open innovation and help foster a self-reliant AI ecosystem. Rajgopal highlighted that, "we are not really dependent in terms of technology," reinforcing the goal of self-reliance.

Open source: The cornerstone of trust and independence

Our discussions at the Open Source Summit underlined that the era of AI competition has moved beyond raw processing power—it is now fundamentally about building trust. This is precisely where open source becomes indispensable.

• Transparency and trust: Open source offers significant transparency, enabling inspection, understanding, and customization of models and software. Open source also gives us the tools for transparency and safety, which are essential for building AI responsibly. This "full 'glass-box' view" allows for verifying model integrity and assessing bias, safety, and fairness for regulatory and ethical compliance. Red Hat's enterprise open source solutions foster trust through transparent development, enhancing security, compliance, and supply chain integrity. Innovation in AI is running ahead of our ability to keep it under control, and open source methods can help solve the trust problem by extending software supply chain principles to AI.
• Flexibility and control: Open source also provides the flexibility and control needed to easily cater to various sovereign requirements. It helps avoid vendor lock-in by offering open standards and a consistent experience across different environments. This way, if your sovereignty requirements evolve, migrating smoothly becomes much simpler.
• Rapid innovation: Open source projects encourage rapid innovation, with open-weight models rapidly catching up to and in some cases surpassing closed source alternatives. Key open source projects like vLLM and Kubernetes, which has become the de facto platform to orchestrate AI workload at scale, are crucial for this innovation, supporting a range of hardware like NVIDIA, AMD, and Intel GPUs.

NxtGen SpeedCloud: India's sovereign AI in action

NxtGen is realizing this vision with SpeedCloud, a sovereign AI platform built on open source technologies. Rajgopal A S described SpeedCloud as going beyond a data center or GPU-as-a-Service to become "a sovereign AI platform built on a foundation of open source technologies."

SpeedCloud’s open architecture enables NxtGen to unify and virtualize various hardware—including NVIDIA, AMD, and Intel GPUs—creating a truly platform-agnostic cloud. This represents a decisive step toward reducing vendor lock-in and establishing genuine technological sovereignty. In this they are advancing beyond simply offering GPU-as-a-Service to delivering a managed, end-to-end AI Platform-as-a-Service (PaaS) for the entire AI lifecycle.

The foundation of NxtGen SpeedCloud relies on key Red Hat technologies:

• Red Hat OpenShift: NxtGen uses OpenShift to abstract the underlying GPU infrastructure and allow customers to focus on their AI use cases. Rajgopal stated, "we are perhaps the only ones who can proudly say that we have a unifying platform which is OpenShift. This abstraction allows the customer to focus on his own AI use case." OpenShift also enables them to host public models to understand infrastructure behavior, scaling, and concurrency.
• Red Hat Enterprise Linux (RHEL) 9: Provides the robust infrastructure for CPU, RAM, storage, NICs, and GPUs. RHEL serves as a consistent, flexible, and security-focused platform that lays the foundation for sovereign cloud environments.
• OpenStack Platform 17.1: Enables scalable compute, networking, and storage services via APIs.
• Cloud Management Platform: Automates resource allocation, deployment, scaling, and optimization, backed by Red Hat's enhanced security, scalability, and management capabilities.
• Red Hat Ansible Automation Platform: Streamlines IT operations through configuration as code, enabling automation and management of Day 1 and Day 2 operations, supporting sovereign strategies through automated lifecycle management.
• Red Hat Ceph Storage: Manages approximately 50 PiB of data, supporting diverse workloads like virtual machines (VMs), FS, S3, backup and recovery, high-performance computing (HPC), and private S3, with features like multisite setup and S3-based data replication for high availability and disaster recovery.

How Red Hat can help

Red Hat helps organizations achieve resilience, autonomy, and independence through our open hybrid cloud approach. Our solutions provide:

• Transparency: Our open source model prioritizes security and reliability, building trust and helping nations manage their digital assets and customize software to their specific requirements.
• Control: Our open hybrid cloud strategy offers extensive deployment options—on-premise, cloud, or edge—and helps avoid vendor lock-in. For example, Red Hat OpenShift simplifies migration across different infrastructures if sovereignty requirements change.
• Operational stability: Red Hat technologies, supported by our partner ecosystem, help protect sensitive data, improve backup and recovery, strengthen business continuity, and enhance software supply chain security, addressing regulations like the EU’s Digital Operational Resilience Act (DORA).

Specifically, Red Hat OpenShift Platform Plus is designed as a zero trust-aligned foundation that embeds digital sovereignty into every layer. It offers features like:

• Policy-as-Code framework: Automates and declaratively governs the application lifecycle, so workloads and data remain confined to authorized regions or compliance zones.
• Continuous validation: OpenShift operators, alongside Red Hat Advanced Cluster Security for Kubernetes and Red Hat Advanced Cluster Management for Kubernetes, continuously validate against configuration drift and policy violations, maintaining alignment with security baselines and sovereignty requirements.
• Comprehensive observability: Provides detailed insights into workload behavior, image security, and access patterns, allowing precise tracking of workload locations and data usage to uphold geographic and regulatory boundaries.
• Data residency, control, and provenance: Through Red Hat Quay's container registry and region-aware workload scheduling, OpenShift enables precise control over data storage locations and access policies.
• Software supply chain security: Red Hat Trusted Software Supply Chain strengthens data integrity by supporting artifact signing, software bills of material (SBOM), and artifact verification, helping protect against hidden backdoors and vulnerabilities.
• Network segmentation: OpenShift enforces zero trust principles at the network level, enabling geo-segmented environments where workloads and services are confined to specific regions or clusters.
• Workload identity and access control: Red Hat's zero trust workload identity manager, based on SPIFFE/SPIRE projects, provides cryptographically-verifiable identities for workloads, enabling protected service-to-service communication.

A shared vision for India's AI future

This collaboration between NxtGen and Red Hat represents a shared vision for India's digital future. Our open source-first approach to AI trustworthiness provides the essential tools and frameworks—from model transparency to supply chain security—for building a responsible AI ecosystem.

By fostering an accessible and sovereign infrastructure, these technologies empower India's startups and researchers to build, test, and deploy AI solutions that meet local needs.

This fusion of India's development-focused AI mission with a scalable, open source platform helps accelerate progress on national priorities, including financial inclusion, agricultural productivity, and education. India’s unique, open source-first approach offers a new framework for AI sovereignty that balances national control with global collaboration, serving as a powerful model for other nations. The rest of the world can take an example from India on how the use of open source has enabled them to fast track AI adoption, as well as helping develop the skills and opportunities for its own people.