The IndiaAI Mission’s compute pillar is building out AI cloud capacity with tens of thousands of NVIDIA GPUs, and NVIDIA named Yotta, Larsen & Toubro and E2E Networks as three of its cloud and infrastructure partners, according to a blog the company published on 17 February 2026. IndiaX24 uses that post as the source for the figures below, so they were current as of February and may have changed since.
Yotta is deploying more than 20,000 NVIDIA Blackwell Ultra GPUs, according to the blog. It runs the Shakti Cloud across campuses in Navi Mumbai and Greater Noida and offers GPU-dense, pay-per-use AI cloud services. A pay-per-use model lets start-ups and researchers rent computing time without buying hardware, which is the reason such capacity matters for smaller companies.
Larsen & Toubro is building what the blog calls gigawatt-scale sovereign AI factories. The initial expansion is 30 megawatts in Chennai, and a new facility of 40 megawatts is planned in Mumbai, according to the blog. The two together come to 70 megawatts, a sum by IndiaX24, which is 7 per cent of one gigawatt. The blog says the sites will host sovereign cloud workloads and hyperscale deployments.
E2E Networks is deploying an NVIDIA Blackwell GPU cluster on its TIR platform, located at the L&T Vyoma Data Center in Chennai, according to the blog. The cluster uses NVIDIA HGX B200 systems and enterprise software. That means that L&T’s Chennai site is the shared location for two of the three partners, an observation by IndiaX24 from the two descriptions.
Netweb Technologies is manufacturing NVIDIA GB200 NVL4 platforms in India, according to the blog, each with four Blackwell GPUs and two Grace CPUs. Making the servers in India matters for the mission’s aim of building domestic capacity, although the blog does not give production volumes.
The mission itself is described as an investment of over 1 billion dollars in compute capacity and sovereign AI development, according to NVIDIA. The blog also says that more than 4,000 Indian AI start-ups are in NVIDIA’s Inception programme, and it lists venture partners including Peak XV, Z47, Elevation Capital, Nexus Venture Partners and Accel India. Those are NVIDIA’s figures and have not been checked against government data.
The numbers raise questions that the blog does not answer. It does not say how many of Yotta’s 20,000 GPUs have been delivered, how much of the capacity is allocated to mission users and at what price, or how much power the Yotta campuses use. A GPU count is a measure of potential and not of use, and IndiaX24 cannot say what share of the capacity is in operation.
The government’s role is in subsidised access. IndiaAI is meant to provide computing to researchers and companies, according to the general description of the mission, but this report does not give the subsidy rate or the list of beneficiaries.
Readers should note that the post is eight months old and was written by a vendor with an interest in presenting the buildout favourably. IndiaX24 includes it because it is the most specific public listing of partner capacity found, and has labelled its date throughout.
IndiaX24 will update the figures if the partners or the IndiaAI Mission publish newer ones.
Image: Server racks in a data centre (file photo; a stand-in image, not an Indian facility). Photo by Carl Lender from Sunrise, USA, licensed under CC BY 2.0, via Wikimedia Commons.
Leave a Reply