Indian Prime Minister Narendra Modi framed India’s AI pitch around scale, openness, and global relevance, speaking at the fourth day of the AI Impact Summit in New Delhi. More than 100 countries including the UAE and Saudi Arabia are represented at the “world’s largest” AI conference.

India’s diversity makes it the toughest proving ground for the technology, Modi said, calling for an open source and shared AI development: “when codes are open and shared, mns of young minds can make them safer and better.” He called out countries and companies that treat AI as a strategic asset that is developed confidentially.

The summit was buzzing with activity as global AI giants announced new partnerships, investments and initiatives in India.

Reliance pledges USD 110 bn

Reliance Industries and its telecom arm Jio will invest USD 110 bn over the next seven years to build AI and data infrastructure, Chairman Mukesh Ambani said (watch, runtime: 14:45) at the summit.

Where the money goes: The spending will focus on long-term “nation-building” assets, including sovereign AI compute capacity, Ambani said, as high costs and limited access to computing power remain the biggest bottlenecks for India’s AI ambitions. Jio is developing multi-GW, AI-ready data centers in Gujarat’s Jamnagar and expects more than 120 MW of capacity to go live in 2H of this year, powered largely by renewable energy.

Why it matters: Ambani’s explicit goal to reduce the cost of AI is similar to what Ambani’s Jio attempted to do with mobile data. For investors and businesses in the MENA-India corridor, the scale of investments from India’s domestic conglomerate trifecta — the Ambani, Adani, and Tata groups — invites the potential for localized, low-latency, and significantly cheaper AI services.

L&T 🤝Nvidia

Mumbai-based engineering major Larsen & Toubro (L&T) announced a JV with American tech giant Nvidia to build a GW-scale AI factory, according to a press release. Under the plan, L&T will develop a large AI data center campus and scale Nvidia GPU clusters to 30 MW at its Chennai facility and 40 MW at a Mumbai site, enabling sensitive data and workloads to remain within India. It adds to Nvidia’s existing AI investment announcements at the India AI Impact Summit.

The venture will support production-scale AI across manufacturing, energy, finance, healthcare, and public services, L&T said. The partnership aligns with India’s ambition to become a global digital infrastructure hub, Nvidia CEO Jensen Huang said.

Why it works: The partnership pairs L&T’s large-scale engineering and infrastructure execution with Nvidia’s full AI stack, spanning GPUs, CPUs, networking, storage, enterprise software, and reference architectures. It also targets sectors facing tightening data residency requirements.

Tata 🤝OpenAI

Tata Consultancy Services (TCS) has signed OpenAI as the anchor customer for its data center arm, according to a press release. Under OpenAI’s global Stargate initiative, the partnership will build AI-ready data center capacity in India, starting with 100 MW and scalable to 1 GW, with a focus on data residency, security, and domestic capability.

The pact is a strategic boost for TCS, which last year outlined plans to invest up to USD 7 bn to build a 1 GW data center platform in India, expanding beyond its core IT services business.

Plus: TCS parent firm Tata Group has partnered with OpenAI to roll out ChatGPT Enterprise across its businesses over the next few years, beginning with thousands of TCS employees. India is ChatGPT’s second-largest market, with over 100 mn weekly users.

Google 🔗India-America

Alphabet CEO Sundar Pichai on Wednesday announced the India-America Connect Initiative, a plan to roll out new subsea cable routes linking the US, India, and parts of the Southern Hemisphere to boost AI-era connectivity, according to a press release. The move comes on the back of Google’s USD 15 bn pledge for a full stack AI hub in southern India and to strengthen high-capacity data flows for AI, cloud services, and cross-border digital trade.

Sarvam AI launches vernacular LLMs

Bengaluru-based startup Sarvam AI unveiled two voice-first large language models at the summit, pitching them as a homegrown alternative to ChatGPT and Claude for India’s linguistically diverse market, Bloomberg reports.

A vernacular vision: The 30B model targets chatbots, translation, and content creation, while the larger 105B model handles complex reasoning tasks, such as analyzing balance sheets in seconds. The models support 22 Indian languages and are designed for voice use, targeting a vast market of vernacular internet users.

Tags: