India AI Summit 2026 news wrap: GPUs, data centers, sovereign AI

AI Summit India 2026 delivered a focused set of enterprise-facing announcements aimed at a single outcome: expanding India-hosted AI capacity and accelerating production deployment across regulated and large-scale environments. Here are some key takeaways clustered around public and private GPU expansion, hyperscale-grade data center plans, major platform partnerships, large capital commitments, and India-language model releases.

National compute capacity: Government adds to IndiaAI GPU pool

The Ministry of Electronics and Information Technology said India’s national AI compute program will expand by adding 20,000 GPUs to an existing base of 38,000 GPUs. The announcement positioned the addition as a near-term step under the IndiaAI Mission framework, intended to broaden access to domestic compute for AI development and deployment.

NVIDIA’s India buildout: Blackwell GPUs land with local providers

US semiconductor company NVIDIA outlined a set of India partnerships tied to its Blackwell generation platforms, with Indian infrastructure operators attaching specific capacity plans to enterprise cloud offerings.

Yotta, a data center and cloud infrastructure provider, announced it will add more than 20,000 NVIDIA Blackwell Ultra GPUs across data center clusters in Navi Mumbai and Greater Noida. Yotta presented the expansion as a pay-per-use AI cloud service designed for enterprise and government workloads.

E2E Networks, a cloud services firm, announced deployments of NVIDIA Blackwell (HGX B200) systems under its “TIR” platform, hosted at a Chennai facility operated by Larsen & Toubro. The announcement positioned the setup as India-hosted advanced GPU access paired with NVIDIA’s enterprise software stack and model tooling, intended to shorten the path from experimentation to production deployments.

NVIDIA also referenced India-based system manufacturing through Netweb Technologies, an IT hardware and systems company, highlighting systems built around NVIDIA’s Grace Blackwell platform as part of the broader push to deepen the local supply chain for AI infrastructure components.

L&T’s sovereign AI infrastructure: Megawatt clusters and gigawatt ambition

Larsen & Toubro, the engineering and construction major, disclosed plans for a proposed venture with NVIDIA aimed at building sovereign AI infrastructure at very large scale. L&T referenced a gigawatt-scale AI data centre campus concept and described planned GPU cluster build-outs at 30 MW in Chennai and 40 MW in Mumbai. The company framed these plans as part of an “AI factory” approach, oriented toward large, dedicated compute campuses intended for sovereign and enterprise workloads.

OpenAI and Tata: Data center capacity deal and internal enterprise rollout

A second major infrastructure storyline came via Tata Consultancy Services, an IT services provider, and OpenAI, an AI company. TCS’s data center unit was reported to have signed up OpenAI as its first customer under an OpenAI infrastructure initiative, beginning at 100 MW and described as potentially scaling to 1 GW. The announcement was presented around enterprise requirements including data residency, security, compliance, and lower latency for mission-critical deployments.

TCS also said it will deploy ChatGPT Enterprise internally, beginning with hundreds of thousands of employees, and described using OpenAI tools to standardize AI-enabled workflows across its engineering and delivery teams. The disclosures positioned the move as both a large internal adoption program and a platform alignment meant to support customer delivery at scale.

Google’s connectivity and infrastructure: Subsea gateway and new routes

Google, the cloud and connectivity provider, announced an India-linked infrastructure program branded “America-India Connect,” framed as part of a multi-year investment push tied to AI infrastructure growth. Google said the plan includes a new international subsea gateway in Visakhapatnam, plus three additional subsea paths connecting India toward Singapore, South Africa, and Australia, along with supporting terrestrial fiber routes. The company presented the initiative as expanding resilient, high-capacity connectivity to support international cloud and AI traffic, alongside broader India infrastructure expansion linked to data and compute.

Big-ticket capital commitments: Reliance, Adani, Microsoft, and Yotta

Several large investment statements were reiterated or unveiled around AI compute and data infrastructure. Reliance Industries, the conglomerate spanning telecom, retail, and energy, said it plans roughly $110 billion of spending over seven years to build AI computing and data infrastructure, describing the effort as focused on addressing compute scarcity and cost constraints.

Adani Group, the infrastructure and energy conglomerate, said it will commit $100 billion toward renewable-energy-powered AI data centres, tying its plan to the electricity requirements of large AI infrastructure and data center buildouts.

Microsoft, the software and cloud provider, reiterated long-horizon investment intent for expanding AI capacity and adoption across the Global South, while referencing previously announced India-related investment commitments over the decade. Yotta also framed its GPU expansion and AI hub buildout as part of a multi-billion-dollar investment program aligned with scaling enterprise-grade AI cloud capacity in India.

Sovereign AI and India-language models: Sarvam AI, BharatGen, and Gnani.ai

The summit also surfaced a set of model launches positioned as sovereign or India-first capabilities intended to broaden local deployment options for enterprises. Sarvam AI, an Indian AI startup focused on language models, launched two large language models trained in India, described as 30B and 105B parameter systems.

BharatGen, a national initiative linked to Indian-language AI development, announced a Param2 17B mixture-of-experts model and said it intends to release the model and workflows as open source.

Gnani.ai, a voice AI company, announced a text-to-speech model designed for voice cloning across 12 Indian languages using short reference audio. The launch was positioned as suited for high-volume service environments such as customer support, contact centers, and public-service delivery, with an emphasis on Indian-language coverage as a core requirement for scaled deployments.

Skills and workforce programs: National skilling and state-level partnerships

On workforce development, the Ministry of Skill Development and Entrepreneurship outlined AI skilling initiatives aligned to readiness, including SOAR (Skilling for AI Readiness) and credentialing pathways linked to the National Skills Qualifications Framework.

At the state level, the Government of Andhra Pradesh, led by Chief Minister N. Chandrababu Naidu, announced an MoU with IBM, the enterprise technology provider, to train one million youth in AI, quantum technology, and cybersecurity. The state also described plans for a Quantum & AI Centre of Excellence and AI “living labs” in partnership with NVIDIA.

The announcement set in one line: capacity, partners, and sovereign options

Across the summit’s enterprise track, the announcements consistently pointed to three deliverables: expanding India-hosted GPU and data center capacity through government and private buildouts, deepening partnerships that bundle platforms with local infrastructure and delivery operators, and launching sovereign and India-language AI models aimed at production deployment needs across large organizations.