Yotta to build $2B AI computing hub using NVIDIA chips
What's the story
Yotta Data Services, an Indian data center company, has announced plans to build one of Asia's largest artificial intelligence (AI) computing hubs. The project will cost over $2 billion and use NVIDIA's latest Blackwell Ultra chips. Under a four-year engagement worth over $1 billion, NVIDIA will anchor one of Asia's largest DGX Cloud clusters within Yotta's infrastructure as part of the project.
Partnership
NVIDIA to deploy GPUs on Yotta's supercluster
NVIDIA will deploy around 10,300 GPUs on Yotta's supercluster for its global Asia-Pacific customers. The move comes as global cloud providers such as Microsoft and Amazon are expanding their AI data center capacity in India. This is in response to the rising demand for generative AI services and a push to localize advanced computing infrastructure.
Capacity expansion
Supercluster to support India's national AI mission
The supercluster, which is expected to be live by August, will be deployed at Yotta's data center campus near New Delhi and its facility in Mumbai. While some GPU capacity will be used for the DGX Cloud supercluster, a big chunk of the remaining capacity will be dedicated to India's national AI Mission. This includes supporting initiatives such as Bhashini, Sarvam, BharatGen and Soket that are building foundational Indian language AI models.
Growth projection
Yotta's GPU footprint set to increase significantly
Yotta's total GPU footprint is set to increase from 40,000 GPUs today to over 75,000 GPUs in the next two years. The company has already raised funds for the entire $2 billion GPU investment and is now planning a pre-IPO and IPO funding round of $1-$1.2 billion. This expansion will significantly boost India's compute capacity by almost five to six times, according to Yotta co-founder Sunil Gupta.