Adani Group Pledges $100B for Green AI Data Centers
Indian conglomerate Adani Group announced it will invest $100 billion by 2035 to build artificial intelligence data center infrastructure powered by renewable energy. The company expects the major initiative to catalyze an additional $150 billion in related investments. The move combines two major global investment trends: the build-out of AI compute capacity and the energy transition.
- The project's energy backbone will be Adani Green Energy's 30 GW Khavda project in Gujarat, which is set to be the world's largest renewable energy plant. More than 10 GW of this project's capacity is already operational. - This initiative will more than double AdaniConneX's (a joint venture with EdgeConneX) data center capacity, scaling from 2 GW to a target of 5 GW. The expansion includes partnerships with Google for a gigawatt-scale AI data center campus and collaborations with Microsoft on projects in multiple cities. - A significant portion of the Graphics Processing Unit (GPU) capacity on the 5 GW platform will be reserved for Indian AI startups, research institutions, and deep-tech innovators to address domestic compute shortages. - Chairman Gautam Adani has stated the goal is for India to be a "creator, builder, and exporter of intelligence," moving beyond being just a consumer in the AI age by building the complete five-layer AI stack. - The investment is part of a broader "sovereign AI" trend, where nations seek to control their own AI capabilities due to geopolitical concerns and data localization mandates. - The data centers will incorporate advanced technologies like liquid cooling and high-efficiency power design to handle the demands of high-density AI computing. - The broader $250 billion ecosystem projection includes catalyzing investments in domestic manufacturing of critical components like servers, high-capacity transformers, and advanced power electronics to reduce supply chain risks. - This move comes as India's data center capacity is projected to double from 0.9 GW in 2023 to around 2 GW by 2026, driven by digitalization and data localization trends.
