India’s $70B AI infrastructure push
A roundup says India’s investment in AI infrastructure has topped $70 billion, combining sovereign and startup spending to build compute capacity and platforms. The coverage frames the flow as infrastructure‑level investment rather than only software funding. (asanify.com)
India says investment tied to artificial intelligence infrastructure has reached about $70 billion, shifting the story from software startups to the physical build-out of computing capacity. (economictimes.indiatimes.com) Electronics and Information Technology Minister Ashwini Vaishnaw put that figure on the record in late January 2026, ahead of the IndiaAI Impact Summit in New Delhi. He said the total reflected spending on computing infrastructure and said the government was preparing IndiaAI Mission 2.0 within five to six months. (telecom.economictimes.indiatimes.com) The public side of that push started in March 2024, when Prime Minister Narendra Modi’s cabinet approved the IndiaAI Mission with a budget of ₹10,371.92 crore, or roughly $1.2 billion at current exchange rates. The plan called for a public artificial intelligence compute platform with 10,000 or more graphics processing units, the specialized chips used to train and run large models. (pmindia.gov.in) By February 2025, the government said the IndiaAI Mission’s common computing facility had expanded to 18,693 graphics processing units. The Press Information Bureau said the program was designed to give startups, researchers, and public agencies access to shared compute rather than forcing each group to build its own server farms. (pib.gov.in) Private capital is arriving at a much larger scale than the state program alone. Microsoft said in December 2025 that it would invest $17.5 billion in India over four years for cloud and artificial intelligence infrastructure, skilling, and operations. (news.microsoft.com) Google said in October 2025 that it would spend about $15 billion over five years on its first artificial intelligence hub in India, centered in Visakhapatnam. Adani Enterprises said the same month that its AdaniConneX venture and Google would build what they called India’s largest artificial intelligence data center campus there, alongside new energy infrastructure. (blog.google) (adani.com) The startup side is smaller in dollar terms but broad in count. Tracxn said in February 2026 that more than 1,700 Indian artificial intelligence-native companies had raised about $5.5 billion in equity funding, with 2025 funding at $856 million and 2026 year-to-date funding at $626 million as of January. (fortuneindia.com) That mix helps explain why officials are calling this an infrastructure build, not just a venture boom. Data centers, graphics processing units, power links, and cloud platforms are long-lived assets that determine where models are trained and where inference, the step where a model answers a user request, actually runs. (pib.gov.in) (news.microsoft.com) India is also trying to reduce dependence on foreign platforms by backing domestic model builders. Government statements on the IndiaAI Mission said the program would support indigenous foundational models, and reports from the 2026 summit said compute allocations had already gone to 12 companies building Indian models. (pmindia.gov.in) (newindianexpress.com) The next test is whether announced money turns into operating capacity on schedule. India has moved from promising shared graphics processing units in March 2024 to counting tens of billions of dollars in announced infrastructure by early 2026, and the government is now using that base to pitch a second phase. (pmindia.gov.in) (economictimes.indiatimes.com)
