India’s sovereign AI push turns to local models

Since its launch in March 2024 with a robust ₹10,371 crore budget, the IndiaAI Mission has spearheaded a strategic and cautious approach to building a sovereign AI ecosystem in India. This initiative emphasizes regulatory oversight and data localization, exemplified by the Ministry of Electronics and Information Technology's temporary halt on deploying OpenAI and Anthropic cybersecurity models to safeguard national interests. Alongside this, the mission integrates comprehensive governance and safety frameworks, including the establishment of the IndiaAI Safety Institute and approval of 13 Responsible AI projects, ensuring that AI deployment aligns with risk-based oversight and ethical standards.

By mid-2026, the IndiaAI Mission had significantly expanded its scope by endorsing 20 indigenous sovereign AI models, such as those developed by Sarvam AI, Gnani.AI, and BharatGen, to reduce foreign dependency and bolster national AI capabilities. This expansion is complemented by a vast network of 58 AI Centres of Excellence and 27 India Data and AI Labs, alongside extensive capacity-building programs like YUVA AI for All, which has trained over 26.5 lakh individuals, reflecting a holistic approach to nurturing both technology and talent within the country.

Beyond AI models and governance, the Indian government has strategically invested in foundational infrastructure through the establishment of 25 Technology Innovation Hubs under the National Mission on Interdisciplinary Cyber-Physical Systems, aimed at accelerating AI technology development and human resource growth. Parallelly, the Semicon India Programme and Semicon 2.0, with combined investments exceeding ₹2.9 lakh crore, underscore a critical policy thrust towards achieving semiconductor self-reliance, a cornerstone for sustaining sovereign AI infrastructure and reducing external vulnerabilities.

NTT DATA's landmark $1.5 billion investment announced in July 2026 underscores a strategic push to expand AI-ready data centers across India, directly supporting the nation's sovereign AI ambitions and stringent data residency mandates. This investment is part of a broader 'hybrid sovereignty' approach that balances local control over sensitive AI data with the benefits of global innovation, enabling resilient and compliant AI deployments particularly suited for APAC economies. Aligning with market forecasts that predict the global sovereign AI sector to reach $148 billion by 2032, with APAC as the fastest-growing region, NTT DATA’s initiative exemplifies the region’s accelerating commitment to sovereign AI infrastructure.

India’s data center capacity has surged more than fourfold since 2020, expanding from 375 MW to 1,575 MW by mid-2026, driven by escalating demand for AI and high-performance computing workloads. This rapid infrastructure growth is complemented by the IndiaAI Mission, a strategic government initiative focused on enhancing compute capacity, developing foundational AI models, and fostering indigenous applications to democratize AI access nationwide. Notably, this expansion is being pursued with a strong emphasis on environmental sustainability, reflecting a conscious effort to balance technological advancement with responsible industry practices.

Industry leaders like SUSE are actively contributing to the sovereign AI ecosystem with innovative infrastructure solutions, exemplified by the August 2026 launch of the 'SUSE AI Factory with NVIDIA' in India. This collaboration integrates NVIDIA AI Enterprise with SUSE’s open-source platforms to provide AI-ready environments that adhere strictly to India’s DPDP Act and CERT-In guidelines, facilitating enterprise adoption of sovereign AI technologies within compliant, private infrastructures. Such initiatives highlight the growing synergy between open-source innovation and regulatory compliance in building robust sovereign AI capabilities.

By mid-2026, India’s indigenous AI ecosystem for national security witnessed significant momentum with startups like Innefu Labs securing $30 million from Panthera Growth Partners to accelerate their proprietary Agentic AI platform and expand AI infrastructure tailored for high-trust environments. This investment, one of the largest in the country’s national security tech sector, underscores the rising demand driven by the Atmanirbhar Bharat initiative and domestic procurement priorities, positioning Innefu for a global expansion and IPO while reinforcing sovereign capabilities without foreign cloud dependencies.

Simultaneously, the Indian government deepened its collaboration with domestic AI firms Sarvam and BharatGen to develop indigenous foundation models aimed at advanced cyber-defence for critical infrastructure sectors such as banking, telecom, energy, and transport. These AI systems are engineered to operate within isolated government networks, ensuring sovereignty and control in sensitive cyber operations, marking a strategic shift towards self-reliant, secure AI deployments that could transform national cybersecurity frameworks.

The IndiaAI Mission’s July 2026 endorsement of 20 sovereign AI model proposals, including Sarvam’s large language models and BharatGen’s multilingual foundation models, further cemented the government’s commitment to reducing foreign technology dependence by fostering indigenous AI platforms tailored for national security and critical sectors. This milestone reflects a coordinated effort to build a robust sovereign AI ecosystem that supports diverse domain-specific needs while safeguarding data sovereignty.

Innefu Labs’ August 2026 launch of Sarvagata AI epitomizes this indigenous AI evolution, delivering an on-premise, ruggedized AI platform designed for India’s national security and sensitive sectors that guarantees data sovereignty through full air-gap deployment. Featuring advanced autonomous capabilities such as multilingual OCR, generative AI, video analytics, and OSINT processing within a modular, encrypted architecture, Sarvagata AI addresses the critical intelligence gap faced by organizations restricted from using public AI systems. As CEO Tarun Wig emphasized, 'Sarvagata AI was built to change that equation permanently,' signaling a transformative leap in securing and modernizing critical workflows.

By mid-2026, India's cybersecurity strategy revealed a pronounced shift towards sovereign AI ecosystems, exemplified by the Ministry of Electronics and Information Technology's advisory halting deployments of foreign AI cybersecurity models from OpenAI and Anthropic. This move underscored deep concerns over dual-use risks, regulatory gaps, and data localization challenges inherent in relying on external AI technologies. Concurrently, CERT-In ramped up national cyber resilience through expansive sector-wide exercises involving 1,470 participants from 345 organizations, deploying AI-driven situational awareness and vulnerability assessments to counter AI-enabled threats, thereby reinforcing a homegrown defense posture.

Complementing these defensive measures, India launched the IndiaAI Safety Institute under the ₹10,371 crore IndiaAI Mission in July 2024, signaling a proactive commitment to mitigating AI risks such as bias, misinformation, and security vulnerabilities within its sovereign AI framework. The institute oversees 13 responsible AI projects focusing on critical areas including bias mitigation, privacy-preserving AI, explainability, deepfake detection, and machine unlearning, reflecting a comprehensive approach to securing AI ecosystems beyond mere technological safeguards.