The ambition for AI in India is moving into a new phase following the rapid growth of AI talent and Data and the slow rate of growth in India’s ability to build out and deploy compute capacity. The inadequate computing infrastructure in the local Indian market has created some significant limitations for the development of local ML Models and their scalability. Yotta Data Services has announced plans to invest $2 billion into building a new AI Hub and utilize NVIDIA GPUs as the basis of this commercial venture.
The company claims that demand for compute resources is already greater than their current ability to supply compute resources because the new Indian AIs are scaling quickly, and acceptance of Indian-made AIs is growing rapidly among users. This project will also allow Yotta to consider a public listing, tying the success of their infrastructure build-out with confidence from capital markets.
What the build-out includes and why scale matters for India AI compute
According to Yotta CEO Dinesh Katiyarkar, their plan calls for the deployment of over 20,000 Nvidia Blackwell Ultra GPUs to support their data centers. Currently, they have around 10,000 GPUs operating, and claim to control approximately 60%-70% of the entire GPU capacity in India. Yotta’s primary sites for expansion will be Greater Noida, enabling the creation of a distributed super-cluster, while also continuing development at Navi Mumbai. Company executives believe the major challenge to achieving this is in delivering inference (the ability to serve up large numbers of applications to users at a global level) and that, in order to do so, they will need to provide users with fast, dependable computing based in proximity to them.
How demand, partners, and IPO plans align
There has been a great deal of momentum throughout the ecosystem recently. For example, the India AI summit had Indian-based models run remotely on aggregate processing power (i.e., GPUs) at the Yotta data centre facility. In addition to that, major global players such as Google and Microsoft have committed billions of dollars to establish data centres in India. OpenAI has agreed with Tata’s data centre subsidiary for data centre capacity.
Yotta is launching a pre-IPO round to raise between $1.2 and $1.5 billion and has a target to go public within the next 12 months. These efforts will help sustain Yotta’s assumed growth, as we project the total amount of data centre capacity in India will be at 1.93 gigawatts in 2025 and could be as much as 4 gigawatts by the end of 2028. This, combined with the establishment of regional AI compute capabilities through these facilities, should position India to be a global leader in this emerging technology segment.
For more- https://civiclens.in/category/https-civiclens-in-technology/
About the Author
