The inner workings of Asia's largest data centre, in Navi Mumbai

Summary

The video explores the rapid growth of India's data center industry, driven by surging data consumption, AI, and the demand for high-performance computing. Focused on Yota Data Services, the Mumbai-based startup backed by Hiranandani Group, the script highlights the company's milestone as the first Indian firm to acquire Nvidia AI chips. Co-founder Sunil Gupta discusses the challenges and opportunities of building AI-specific data centers and the country's role in the global AI market. With a $1 billion investment in cutting-edge technology, Yota aims to cater to diverse industries and address the increasing need for AI infrastructure in India.

Takeaways

• 😀 India has over 820 million active internet users, contributing to a surge in data consumption and the need for more data centers.
• 😀 The demand for data centers in India is rapidly increasing, with the country housing 151 data centers, ranked 14th globally.
• 😀 Yota Data Services, a Mumbai-based startup, is expanding its footprint with backing from the Hiranandani Group.
• 😀 Yota became the first Indian company to acquire AI chips from Nvidia, marking a significant milestone in AI hardware acquisition.
• 😀 Yota's data center in Navi Mumbai is equipped with 7,200 server racks, drawing power from almost 60 megawatts of energy.
• 😀 Sunil Gupta, co-founder of Yota, emphasizes the importance of India creating its own AI models to solve local problems and lead globally in AI development.
• 😀 Nvidia has provided Yota with high-priority allocations of 16,000 powerful Nvidia H100 chips, with plans to build one of the largest supercomputers globally.
• 😀 Yota's purchase of AI chips, worth nearly $1 billion, requires substantial infrastructure, including high-capacity power, cooling, and specialized equipment.
• 😀 Before generative AI's rise, data centers primarily served non-AI workloads, with CPUs handling tasks such as enterprise services and website hosting.
• 😀 The data center industry is witnessing a shift towards AI-specific workloads, with companies increasingly focusing on the infrastructure needed for generative AI tasks.
• 😀 The Indian data center industry has grown rapidly, expanding from 200 MW in the late 1990s to over 1 GW today, with projections showing further growth to 3 GW by 2030.

Q & A

What is the current state of the data center industry in India?

India has over 820 million active internet users, and the demand for data centers is rising rapidly. India ranks 14th in the world in terms of the number of data centers, with 151 centers. However, the demand for data centers is far outpacing the supply, leading to fast market growth and significant investments in this sector.

Why has the demand for data centers increased in India?

The demand for data centers has surged due to the rise of artificial intelligence (AI), the Internet of Things (IoT), machine learning, and other emerging technologies. These factors have greatly increased data generation and consumption, making data localization crucial.

What role does Yota Data Services play in the Indian data center market?

Yota Data Services, a Mumbai-based startup backed by the Hiranandani Group, is rapidly expanding its footprint in India’s data center market. The company became the first Indian firm to acquire AI chips from Nvidia, positioning itself as a leader in providing AI-centric data services.

How has Yota contributed to the development of AI in India?

Yota took a significant step in advancing AI in India by acquiring Nvidia’s AI chips. These chips are critical for training AI models. Yota is working towards becoming a key player in the AI space, not just in India but globally, by offering AI-specific infrastructure and services.

What is the significance of Nvidia's AI chips to Yota's operations?

Nvidia's AI chips are crucial for Yota’s AI infrastructure. Yota acquired 16,000 H100 chips, making it a major provider of AI processing power in India. These chips will be used to build one of the world's largest supercomputers, enabling Yota to handle substantial AI workloads, such as training large language models like GPT-4.

What are the challenges in building AI-specific data centers?

Building AI-specific data centers requires extensive customization, including specialized infrastructure, high power requirements, cooling systems, and advanced networking. The data centers must also support advanced AI chips, which cannot be hosted in regular office environments due to the sheer power and cooling demands.

How does Yota plan to scale its operations with the Nvidia chips?

Yota plans to scale its operations by deploying the 16,000 Nvidia H100 chips in its data centers, which will be interconnected via an InfiniBand network. This infrastructure will be powerful enough to support the AI needs of multiple customers simultaneously, including large-scale AI model training.

What is the projected growth of the data center market in India?

The data center market in India is expected to grow significantly, with projections indicating a shift from 1 gigawatt (GW) of capacity to about 3 GW by 2030. This growth is driven by the increasing demand for cloud services, AI workloads, and the digital transformation across industries.

How is AI impacting the data center industry globally and in India?

AI is creating a new wave of demand for data centers, specifically for AI workloads. The capacity needed to handle AI workloads may soon match the total capacity required for cloud services globally. In India, this is leading to a surge in data center investments and infrastructure development to meet the growing demand.

What types of customers does Yota serve in its data centers?

Yota serves a wide range of customers, including enterprises, startups, banking and financial services, pharma companies, e-commerce businesses, and cloud service providers. These customers consume various services, from power, space, and cooling to hardware, software, virtualization, and cybersecurity.