Yotta Data Services has unveiled plans to build one of Asia's largest AI superclusters in India, investing over US$2 billion (S$2.5 billion) to deploy 20,736 Nvidia Blackwell Ultra GPUs. The infrastructure, aimed at boosting India's AI capabilities, is expected to go live by August 2026.
The supercluster will be hosted at Yotta’s 60 MW D2 hyperscale data centre in Greater Noida, with the campus scalable to 250 MW. Additional support will come from its Navi Mumbai data centre campus, scalable up to 2 GW. The sites feature integrated high-voltage substations, dedicated power distribution and green energy sourcing.
In parallel, NVIDIA has announced plans to establish one of the largest NVIDIA DGX Cloud clusters in the Asia-Pacific region within Yotta’s HGX B300 Blackwell Ultra environment. The four-year engagement, valued at over US$1 billion, builds on an existing relationship under which NVIDIA DGX Cloud has been using Yotta’s GPU infrastructure over the past year.
Sunil Gupta, Co-Founder, MD & CEO, Yotta Data Services, added, “India’s AI ambition requires sustained, high-performance compute at scale. By combining Blackwell Ultra infrastructure with open models like NVIDIA Nemotron and the full NVIDIA AI stack, we are enabling developers to build sovereign, globally competitive AI applications from India.”
Built for trillion-parameter AI models
Yotta’s Blackwell-based supercluster is designed using NVIDIA reference architecture and integrates 800 Gbps NVIDIA Quantum-X800 InfiniBand networking, liquid-cooling technologies and more than 40 petabytes of high-performance parallel file-system storage.
The infrastructure is engineered to support trillion-parameter foundation model training and large-scale inference workloads capable of handling millions of simultaneous prompts.
Beyond compute capacity, Yotta is enhancing its Shakti Studio AI platform with NVIDIA Nemotron open models, NVIDIA NIM microservices and the broader NVIDIA AI Enterprise software stack. Through this integration, the company said, developers in India will gain access to open model weights, datasets and training recipes, enabling fine-tuning and sovereign AI development within domestic infrastructure.
Alongside the DGX Cloud deployment, Yotta has also committed more than 10,000 NVIDIA B300 GPUs from the supercluster to the IndiaAI Mission. The allocation is intended to support Indian foundation model development, research institutions, startups and public-scale AI initiatives, ensuring that domestic AI priorities advance alongside global capacity expansion.
For India’s AI ecosystem, access to large-scale Blackwell infrastructure within the country reduces reliance on offshore compute resources and enables enterprises to scale AI solutions locally.
Yotta currently operates over 10,000 NVIDIA GPUs in production, with another 8,000 scheduled to go live in the coming months. Following the Blackwell Ultra deployment, the company plans to scale beyond 80,000 GPUs by 2027, with long-term ambitions of exceeding one million GPUs over the next three to five years.
