India builds local models and computing capacity

India is expanding its role from a major AI market to a developer of domestic technology. The IndiaAI Mission, launched in March 2024 with a five-year outlay of Rs 10,371 crore, subsidises GPU access and supports home-grown foundation models. Its priorities include linguistic inclusion, data sovereignty and affordable deployment.

In November 2026, the mission presented Soket Labs’ EKA-1 and ZenteiQ AiTech Innovations’ Dhīr 2B and Dhīr 5B. EKA-1 targets maths, coding, reasoning and agentic tasks, with reported support for all 22 scheduled Indian languages. ZenteiQ describes its smaller models as open source, supporting English and 10 Indian languages, with reported offline mobile operation. Their potential lies in local applications rather than direct competition with frontier systems.

Infrastructure proposals include a Telangana AI data centre campus with potential capacity of one gigawatt and investment of up to $7.4 billion. Reliable power, water and suitable locations remain critical conditions. Enterprise adoption is also advancing: TCS reports annualised AI revenue of over $2.3 billion as it moves toward agent-led business workflows, leaving employment implications unresolved.

Through the AI Impact Summit 2026, India promoted inclusive, trustworthy and energy-efficient AI focused on Global South needs. However, limited computing access, gaps in Indian-language training data, talent retention and immature regulation remain obstacles. Independent benchmarking, sustained funding and actual adoption will determine whether announcements translate into durable capabilities.