India’s AI Impact Summit concluded this week with over $250 billion in infrastructure commitments, positioning India as the world’s second AI superpower. The five-day event (February 18-23) drew delegations from 100+ countries, 20+ heads of state, and top AI CEOs including OpenAI‘s Sam Altman and Anthropic‘s Dario Amodei. The summit ended with 89 countries signing the “New Delhi Declaration” – a geopolitical statement challenging Western AI dominance.
Moreover, this isn’t just market expansion. With 100 million weekly ChatGPT users in India (second only to the US), $210 billion from Adani and Reliance alone for data centers, and new open-source models supporting 22 languages, India is building an alternative AI ecosystem. For developers, this means new job opportunities, certification programs outside the US for the first time, and a choice between closed Western models and open Indian alternatives.
OpenAI’s $500B Stargate Expands to India
OpenAI partnered with Tata Consultancy Services (TCS) to deploy 100 megawatts of data center capacity through TCS HyperVault, with plans to scale to 1 gigawatt as part of the company’s $500 billion global Stargate initiative. Sam Altman confirmed India has “more than 100 million weekly active ChatGPT users, second only to the U.S.” – a market too large to ignore.
The partnership extends beyond infrastructure. OpenAI announced 100,000+ ChatGPT Education licenses across Indian universities including IIM Ahmedabad and AIIMS Delhi. TCS becomes the first non-US organization in OpenAI’s certification program, opening AI credentials to developers outside America for the first time. TCS will deploy ChatGPT Enterprise to hundreds of thousands of employees, ranking among the largest enterprise AI rollouts globally. The company will also integrate OpenAI Codex across its engineering teams.
Consequently, OpenAI is opening new offices in Mumbai and Bengaluru beyond its existing New Delhi presence. The local data centers solve India’s data residency requirements for regulated sectors like banking, healthcare, and government – reducing latency from 200-300ms to under 50ms and making OpenAI viable for production deployments that previously couldn’t leave Indian borders.
Anthropic Bets on India AI Summit with Bengaluru Office
Anthropic opened its Bengaluru office this week – its second in Asia after Tokyo – and announced a partnership with Infosys to deploy Claude-powered AI agents across telecommunications, financial services, manufacturing, and software development. India is Anthropic’s second-largest market globally, with CEO Dario Amodei noting that “nearly 50% of Indian Claude usage is tied to building applications, modernizing systems, and shipping production software.” In fact, India isn’t just consuming AI – it’s building with it at scale.
The Infosys collaboration focuses on “agentic AI” – systems that independently handle multi-step tasks rather than just answering questions. Think insurance claims processing, automated code generation and testing, or compliance reviews that run autonomously. Infosys is establishing a dedicated Anthropic Center of Excellence using the Claude Agent SDK. The center starts with telecommunications before expanding to other regulated industries.
This mainstream arrival signals a shift for developers. Agentic AI skills – building multi-step workflows, managing long-running tasks, implementing tool use – are no longer experimental. They’re becoming standard requirements as companies like Infosys hire aggressively for these roles.
$210B India AI Infrastructure: Adani and Reliance Bet
Indian conglomerates Adani Group and Reliance Industries committed a combined $210 billion to AI and data infrastructure – the largest private sector investment announcement at the India AI Summit. Adani pledged $100 billion to expand data centers from 2GW to 5GW by 2035. An additional $150 billion will flow into manufacturing, infrastructure, and sovereign cloud platforms. Reliance matched this commitment with $110 billion, bringing 120 megawatts online in H2 2026 with a path to gigawatt-scale compute.
Notably, the scale rivals existing US data center capacity. Adani’s 5GW target alone would make India a major AI infrastructure hub, not just a consumer market. The Indian government added a $1.1 billion VC fund specifically for AI and advanced manufacturing startups, showing coordinated public-private investment rarely seen at this scale.
As a result, this transforms the economics of building AI applications for developers and startups. Indian infrastructure costs 30-50% less than US equivalents, subsidized compute becomes available through government programs like the IndiaAI Mission, and data sovereignty shifts from compliance burden to competitive advantage. Indian companies can now sell into regulated sectors by emphasizing local data processing – something foreign competitors can’t match without similar infrastructure investments.
Open-Source Counteroffensive: Indian Models Challenge GPT
While partnering with OpenAI and Anthropic, India launched its sovereignty play. Indian AI startups Sarvam AI and BharatGen released major open-source models at the India AI Summit. Sarvam launched Sarvam-30B (optimized for real-time conversation, trained on 16 trillion tokens with 32K context) and Sarvam-105B (designed for complex reasoning with 128K context window). BharatGen unveiled Param 2, a 17 billion parameter model supporting 22 Indian languages. Importantly, all models are fully open-sourced and trained entirely in India.
Specifically, Sarvam’s 30B model uses a mixture-of-experts architecture for low-latency applications, while the 105B model targets code generation and long-form content. Both are available on Hugging Face for free deployment. Sarvam announced partnerships with Qualcomm, HMD, and Bosch to deploy models on edge devices – bringing AI to smartphones and IoT rather than just cloud infrastructure.
Ultimately, this is India’s hedge. Even while welcoming OpenAI and Anthropic, the country is building open-source alternatives optimized for Indic languages and local deployment. For developers, this means free alternatives to GPT and Claude, especially valuable for applications requiring multilingual support (Hindi, Tamil, Bengali, etc.) or cost-sensitive high-volume scenarios. However, the open question is whether these models can match GPT-4 or Claude 3.5 quality in 2-3 years. If they can, India builds an AI ecosystem independent of Western companies. If they can’t, at least the country tried while the infrastructure buildout continues.
The Geopolitical Play: India as AI’s Third Pole
The India AI Summit concluded with 89 countries and international organizations – including both the United States and China – signing the “New Delhi Declaration on AI Impact.” The nonbinding agreement emphasizes democratizing AI access, expanding applications in healthcare and education, and ensuring ethical safeguards. It was rooted in the Sanskrit principle “Sarvajan Hitaya, Sarvajan Sukhaya” (welfare for all), positioning India as a bridge-builder between the US/EU and the Global South.
Significantly, the fact that both the US and China endorsed the declaration reveals India’s strategic positioning. The country is building neither a US-aligned nor China-aligned AI ecosystem, but a third pole. India formally joined the Pax Silica Initiative, signaling a decisive geopolitical shift. Nevertheless, the declaration sidestepped a critical reality: computing power, training data, and frontier AI model capabilities remain concentrated in a handful of Western companies.
Still, India can sign agreements and build data centers, but whether it can execute on these $250 billion commitments is the real question. The country’s track record on infrastructure promises is mixed – remember the Smart Cities Mission? However, this time private sector giants and foreign AI companies have skin in the game, which adds accountability previous government initiatives lacked.
What This Means for Developers
India’s transformation from AI consumer to AI infrastructure hub creates tangible opportunities. OpenAI certifications are now accessible via TCS outside the US. Infosys and other IT giants are hiring aggressively for AI agent development roles. The 100,000+ ChatGPT Education licenses create a pipeline of AI-trained talent entering the job market. Furthermore, open-source models like Sarvam and Param 2 offer free alternatives for developers building cost-sensitive or multilingual applications.
The strategic question for developers: which ecosystem to bet on? Closed Western models (GPT-4, Claude 3.5) still lead on quality but lock you into expensive APIs and foreign data centers. In contrast, open Indian models offer sovereignty and zero API costs but currently trail on benchmarks. The smart play might be hybrid – use open models for high-volume, cost-sensitive tasks and fall back to closed models for edge cases requiring highest quality.
Ultimately, watch the execution. $250 billion in commitments doesn’t guarantee $250 billion deployed. Infrastructure buildouts take years, and geopolitical risks remain – if US-India relations sour, OpenAI and Anthropic could exit like Western companies left Russia. However, if even half these commitments materialize by 2030, India’s AI ecosystem becomes a legitimate alternative to US dominance. That’s a shift developers can’t afford to ignore.
