NVIDIA has just reportedly secured massive AI GPU orders from Indian data center operator Yotta, which sees them steering the AI GPU business away from China and into the arms of India.
The news is coming from Yotta's chief executive, who talked with reporters at Reuters that the company was purchasing more AI GPUs from NVIDIA worth $500 million, taking its total order book with the AI GPU leader to $1 billion. Yotta is beefing up its AI cloud services and needs more AI GPUs to power it.
Yotta CEO and co-founder Sunil Gupta told Reuters that his company would feature close to 16,000 of NVIDIA's AI GPU chips spread between the H100 and upcoming GH200 AI GPUs and will be placed by March 2025. Last year, Yotta placed an order with NVIDIA for close to 16,000 x H100 AI GPUs that are due for delivery to Yotta by July 2024.
NVIDIA secured other AI partnerships with Indian conglomerates Reliance Industries and Tata Group to develop cloud infrastructure and large language models (LLMs) for generative AI applications. As for Yotta, they're part of Indian billionaire Niranjan Hiranandani's real estate group and are a partner firm for NVIDIA in India that has hardware inside of three data center campuses: Mumbai, Gujarat, and near New Delhi.
It was only a month ago that Tianfeng Securities analyst Ming-Chi Kuo said that NVIDIA plans on boosting the volumes of AI GPU shipments in 2024; here's what the analyst expects to see in 2024:
- NVIDIA AI server shipments are expected to grow 150% YoY or more in 2024, and ultra-low loss CCL is currently in tight supply. In addition to existing high-end general servers and PCs/laptops, future AI PCs will also need to upgrade their CCL to ultra-low consumption.
- The CCL usage of AI servers is about 8 times that of general servers. NVIDIA AI servers are expected to further increase CCL usage after they are upgraded to the B100 solution in 2H24.