New Delhi/Singapore, March 4, 2026 — Indian data center and cloud services provider Yotta Data Services has unveiled plans to invest more than $2 billion in deploying 20,736 liquid-cooled Nvidia Blackwell Ultra GPUs, creating one of Asia’s largest AI computing superclusters. The ambitious project, announced during the India AI Impact Summit in February, is set to go live by August 2026 and marks a major step in addressing India’s surging demand for high-performance AI infrastructure.
The supercluster will be hosted primarily at Yotta’s 60 MW D2 hyperscale data center in Greater Noida (near New Delhi), with the campus scalable to 250 MW. Additional capacity will come from Yotta’s Navi Mumbai facility, which has potential to expand to 2 GW. The deployment features Nvidia’s latest HGX B300 Blackwell Ultra architecture with high-speed 800 Gbps Quantum-X800 InfiniBand interconnects, optimized for training and inference of frontier-scale AI models.
As part of a multi-year strategic partnership valued at over $1 billion, Nvidia will establish one of the Asia-Pacific region’s largest DGX Cloud clusters within Yotta’s infrastructure. Nvidia DGX Cloud has already been leveraging Yotta’s GPU resources over the past year, and this expansion significantly scales the collaboration to meet growing regional and global demand.
Yotta CEO Sunil Gupta, speaking to Reuters on the sidelines of the summit, highlighted the acute GPU shortage in India: “Demand for GPUs far exceeds supply as domestic AI models scale and enterprise adoption accelerates. This Blackwell Ultra deployment will help position India among a select group of countries capable of hosting frontier-scale AI infrastructure.”
The initiative directly supports India’s national IndiaAI Mission, with expectations that a substantial portion of the compute—over 10,000 GPUs—will contribute to sovereign large language model development. It also strengthens India-U.S. technology ties, aligning with broader commitments made at the summit by global tech leaders.
Yotta is in the pre-IPO phase and plans to raise up to $1.2 billion from high-net-worth individuals, family offices, and strategic investors to fund further expansion. Sources familiar with the matter told Reuters that Abu Dhabi’s sovereign wealth fund Mubadala is in discussions to invest in the company.
The announcement, first exclusively detailed by Reuters on February 18, 2026, quickly drew follow-up coverage from Bloomberg, CNBC International, The Economic Times, HPCwire, Data Center Dynamics, and others. CNBC’s February 27 feature on its Inside India program included an in-depth interview with Gupta, who noted that “the majority of Indian-developed AI models are currently trained on Yotta-hosted Nvidia GPUs.”
Industry observers view the project as a catalyst for India’s data center boom. With domestic AI startups, large IT services firms (such as TCS, Infosys, and Wipro), and international clients seeking high-density AI cloud capacity, the supercluster is expected to attract significant enterprise adoption once operational.
Nvidia’s deepening footprint in India—through partnerships with Yotta, E2E Networks, and others—underscores the country’s rising importance in the global AI supply chain. Analysts anticipate accelerated investment in Indian AI infrastructure through 2026–2027 as GPU supply constraints ease and national policy support intensifies.
Yotta stated that priority access will be given to Indian AI innovators and global enterprises requiring cutting-edge compute. Further updates on financing milestones, customer wins, and performance benchmarks are anticipated in the coming months.
(Sources: Reuters, CNBC International, HPCwire, Yotta official statements, and related coverage. All figures and timelines based on public disclosures as of early March 2026.)