New Delhi: India has marked a defining moment in its journey toward technological independence by launching three powerful homegrown artificial intelligence systems during the India AI Impact Summit 2026. Held at Bharat Mandapam from February 16–20, the landmark event brought together global leaders, technology executives, researchers, and innovators to chart the course for responsible, inclusive, and sovereign AI development.
Prime Minister Narendra Modi and Union Minister Ashwini Vaishnaw used the platform to underline India’s vision of AI that serves humanity, respects cultural diversity, protects national interests, and remains firmly under domestic control.
Prime Minister Modi’s Call for Ethical and Inclusive AI
In a widely anticipated address at the Leaders’ Plenary Session on February 19, Prime Minister Narendra Modi described artificial intelligence as one of the most consequential technologies of our era. He warned that its direction will determine whether it becomes a force for universal progress or widens existing inequalities.
Drawing inspiration from Lord Buddha’s principle that right understanding must precede right action, the Prime Minister urged the international community to act decisively and ethically at this critical juncture.
He presented his M.A.N.A.V. framework for guiding AI development:
- Morality — grounding systems in strong ethical foundations
- Accessibility — ensuring benefits reach every citizen
- National Sovereignty — maintaining control over data, models, and infrastructure
- Augmentation — using AI to amplify human potential
- Values — embedding timeless human principles into every system
Modi highlighted successful Indian examples of technology serving people — the world’s largest digital vaccination campaign and the revolutionary Unified Payments Interface (UPI) — as models for how AI should function: empowering rather than dominating.
He cautioned against the dangers of unchecked optimization, famously illustrated by the “paper clip problem,” where a system pursuing a single narrow goal could produce catastrophic unintended outcomes. To counter such risks, he advocated for ethical boundaries that are as expansive as the capabilities of modern AI.
The Prime Minister proposed three foundational global principles for trustworthy AI:
- A secure, sovereign-respecting global data framework that guarantees high-quality, unbiased training inputs
- Transparent “glass box” safety mechanisms that allow verification instead of opaque “black box” decision-making
- Explicit incorporation of human values to guide AI behavior and prevent harm
He concluded by reaffirming India’s determination to center the priorities of the Global South in AI governance and to keep the technology open, collaborative, and people-first.
New Delhi Frontier AI Impact Commitments Launched
Union Minister Ashwini Vaishnaw formally announced the New Delhi Frontier AI Impact Commitments during the summit’s opening ceremony. This voluntary coalition unites Indian innovators — Sarvam AI, BharatGen, Gnani.ai, and Soket — with leading global frontier AI organizations.
The commitments concentrate on two strategic priorities:
- Collecting anonymized real-world usage data to inform evidence-based policies on employment transformation, skill evolution, productivity improvement, and economic restructuring
- Building stronger multilingual and culturally contextual evaluation frameworks, including datasets, benchmarks, and expertise tailored to underrepresented languages and societies, particularly across the Global South
Minister Vaishnaw emphasized that India is pursuing a holistic strategy across every layer of the AI stack — compute infrastructure, foundational models, applications, talent development, and sustainable energy — while always placing human safety, dignity, and inclusion at the core.
Three Breakthrough Sovereign AI Models Unveiled
The summit’s most anticipated announcements were the public debuts of three advanced AI systems developed entirely within India under the IndiaAI Mission framework.
Sarvam AI Delivers Voice-First Multilingual LLMs
Bengaluru-based Sarvam AI introduced two large language models purpose-built for Indian users. These systems support natural voice interactions in 22 Indian languages, addressing the reality that the majority of India’s 1.45 billion citizens are far more comfortable communicating in regional languages than in English.
Developed and hosted completely on Indian soil, the models reportedly achieve superior performance compared to certain global benchmarks set by systems such as DeepSeek R1 and Gemini Flash in relevant Indian-use evaluations. Sarvam positions itself as “sovereign by design,” giving users full confidence that their data and interactions remain under national control.
Gnani.ai Unveils Voice Cloning and Speech Recognition Powerhouse
Also from Bengaluru, Gnani.ai launched its Vachana text-to-speech platform, capable of generating highly natural human-like voices in 12 Indian languages using audio samples as brief as 10 seconds.
The system is specially tuned to handle India’s complex linguistic environment — regional accents, frequent code-switching between languages, colloquial expressions, and everyday conversational flow. Gnani.ai simultaneously introduced its complementary speech-to-text model, trained on more than one million hours of authentic Indian audio, establishing it as a robust enterprise-grade solution for voice-enabled services, customer support, accessibility applications, and digital inclusion.
BharatGen Presents Param2 17B Multilingual Mixture-of-Experts Model
BharatGen, India’s pioneering government-supported initiative for sovereign foundational AI, released Param2 17B, a 17-billion-parameter Mixture-of-Experts model engineered to reflect India’s extraordinary linguistic and cultural diversity.
Developed through close technical collaboration with NVIDIA, Param2 17B forms part of a broader ecosystem strategy that connects startups, system integrators, government departments, and venture capital to build targeted AI applications for public governance, national security, agriculture, and commercial sectors.
IndiaAI Mission: Record-Breaking Compute Expansion
All three models were made possible by the IndiaAI Mission, launched in 2024 with a five-year budget of ₹10,371.92 crore.
Originally aiming for 10,000 GPUs, the mission has already deployed 38,000 GPUs available at subsidized rates to researchers, startups, and developers. Minister Vaishnaw announced an immediate addition of 20,000 more GPUs, with further procurement underway to push total capacity well beyond 50,000 units in the near term.
The mission also operates AIKosh, the national dataset platform, which has already released more than 7,500 datasets and 270 AI models as shared national resources.
Looking Ahead: India’s Realistic Path in Global AI
While India may not yet match the raw scale of frontier models developed in the United States or China, analysts recognize the country’s unique advantages: a gigantic domestic market, exceptional talent pool, cost-effective infrastructure, deep public digital systems, and unwavering policy focus on inclusion and sovereignty.
By building AI that truly understands Indian languages, accents, contexts, and values, India aims to become the world’s largest arena for meaningful, large-scale AI adoption — while simultaneously offering the Global South a credible alternative to externally controlled platforms.
The India AI Impact Summit 2026 has sent a clear message: India is not merely adopting artificial intelligence — it is actively shaping a future where AI belongs to its people, reflects its diversity, and advances its national interests.
FAQs
1. What exactly is Sovereign AI, and why is it a big focus for India right now?
Sovereign AI means a country can independently create, train, host, and govern artificial intelligence technologies using its own infrastructure, local datasets, domestic talent, and rules—without depending heavily on foreign companies or platforms.
For India, this approach ensures control over sensitive data, reduces risks from external biases (especially Western-centric ones in global models), supports 22+ Indian languages and regional accents, and aligns AI with national priorities like privacy, inclusion, and cultural relevance. At the summit, leaders stressed that true sovereignty prioritizes secure, locally managed systems over simply copying the largest foreign models. This push helps India serve its 1.45 billion people more effectively while contributing balanced alternatives for the Global South.
2. Which three sovereign AI models were unveiled at the India AI Impact Summit 2026?
Three major homegrown systems were launched or highlighted prominently:
• Sarvam AI (Bengaluru-based startup): Released two large language models (LLMs) built from scratch in India. These support voice commands and work fluently across 22 Indian languages. They are optimized for Indian contexts and reportedly outperform some global systems (like certain versions of DeepSeek R1 and Gemini Flash) on relevant benchmarks. The company emphasizes full domestic control over development, deployment, and data handling.
• Gnani.ai (also Bengaluru-based): Introduced the Vachana text-to-speech (TTS) model, which clones realistic human voices in 12 Indian languages using very short audio samples (as little as 10 seconds). It handles diverse accents, language mixing, and natural everyday speech. They also showcased a complementary speech-to-text (STT) model trained on over one million hours of real Indian audio data, aimed at enterprise voice services, customer support, and accessibility.
• BharatGen (government-supported national initiative, led by institutions like IIT Bombay): Launched Param2 17B, a 17-billion-parameter multilingual Mixture-of-Experts (MoE) foundational model. Developed with NVIDIA collaboration, it focuses on India’s linguistic and cultural diversity and supports applications in governance, agriculture, defense, healthcare, and more. It is part of the first major government-backed effort to build sovereign foundational models tailored for Indian needs.
These models represent a shift toward India-specific, multilingual, and voice-enabled AI that reduces reliance on foreign platforms.
3. What is the IndiaAI Mission, and what are its latest compute capacity updates?
The IndiaAI Mission is a flagship government program launched in 2024 by the Ministry of Electronics and Information Technology (MeitY) with a ₹10,371.92 crore budget over five years. Its main goals include building massive shared AI compute power, creating national datasets (via AIKosh, which has already shared over 7,500 datasets and 270 models), developing applications, upskilling talent, funding startups, and ensuring safe & trusted AI frameworks.
Key recent updates from the summit:
• It originally targeted 10,000 GPUs but has already deployed 38,000 GPUs available at subsidized rates (around ₹65 per hour) to startups, researchers, and innovators.
• Union Minister Ashwini Vaishnaw announced an immediate addition of 20,000 more GPUs, with orders to be placed soon and deployment expected over the next six months.
• This expansion is part of what officials call “IndiaAI Mission 2.0,” focusing on even larger scale, R&D, innovation, and responsible deployment. Partnerships with NVIDIA and domestic providers (like Yotta’s Shakti Cloud with 20,000+ Blackwell GPUs and Larsen & Toubro’s gigawatt-scale AI factories) are accelerating sovereign infrastructure growth.
4. What were the main highlights from Prime Minister Narendra Modi’s speech at the summit?
PM Modi addressed the Leaders’ Plenary Session on February 19, 2026, describing AI as a technology that can elevate humanity if guided correctly. He introduced the M.A.N.A.V. vision for AI:
- Morality → Ethical foundations
- Accessibility → Benefits for everyone
- National Sovereignty → Control over data and systems
- Augmentation → Enhancing human capabilities
- Values → Embedding human principles
He proposed three global measures for responsible AI:
- A trusted data framework respecting sovereignty and quality (“garbage in, garbage out”).
- Transparent “glass box” safety rules instead of hidden “black box” operations.
- Clear human values to avoid risks like the “paper clip problem” (unintended harm from narrow optimization).
Modi emphasized AI as a shared resource, open code, and service to humanity (not domination), citing India’s COVID-era digital successes (vaccination platform and UPI). He stressed prioritizing the Global South and ensuring AI fosters inclusion without deepening divides.
5. What are the New Delhi Frontier AI Commitments, and who is involved?
Announced by Minister Ashwini Vaishnaw at the summit’s opening, these are voluntary commitments uniting Indian innovators (Sarvam AI, BharatGen, Gnani.ai, Soket) with global frontier AI companies.
The two core focus areas are:
- Real-world AI usage insights → Sharing anonymized, aggregated data on how AI affects jobs, skills, productivity, and economies to guide better policymaking.
- Multilingual and contextual evaluations → Developing datasets, benchmarks, and expertise for underrepresented languages and cultures (especially in the Global South) to improve AI performance and access in diverse settings.
The commitments aim to promote equitable, culturally sensitive AI deployment while allowing flexibility in tools and methods. They reflect India’s strategy to work across the full AI stack (compute, models, applications, talent, energy) with human safety and dignity at the center.
