AI and Cloud demand drive data center growth in India, says PDG's MD Vipin Shirsat

India's data center market is growing quickly, fueled by more digital adoptions, the rise of artificial intelligence (AI), and a shift to cloud computing. The market is expected to reach about $4.5 billion by 2025, according to a report published by market research firm Astute Analytica in July, backed by strong investments and government support for infrastructure. 

Princeton Digital Group (PDG), based in Singapore, focuses on providing scalable data center solutions in Asia. Founded in 2017, PDG has expanded into key markets like India, Indonesia, and Malaysia to meet the rising demand for cloud services. 

In an interview with TechCircle, Vipin Shirsat, Managing Director of PDG, discussed the factors driving data center growth in India, the impact of AI, and the company's recent $1 billion investment in India as part of its $5 billion plan for AI-ready data centers in Asia. Edited Excerpts:  
 
What are the key factors that are driving the data centre growth in India? 

Over the last decade, enterprise cloud has been the main driver for the data center industry worldwide, including in India. This has been the steady growth story for data centers everywhere. However, in the past two years, AI has emerged as a new growth driver globally. 

Major AI technology providers are now forming their infrastructure strategies and targeting specific markets across Asia. These markets are seeing significant growth in demand for data center capacity. Key locations include Johor in Malaysia, Tokyo in Japan, parts of Australia like Melbourne, and importantly, Mumbai in India, which is seeing a significant increase in data center development.

In recent years, hyperscalers have started building their own data centers, particularly in cities like Hyderabad and Pune. These centers are being developed to support both enterprise cloud and AI capacity, leading to an AI spillover effect in these regions. Therefore, the recent surge in demand for data centers in India, especially in Mumbai, is largely driven by AI, and similar growth is observed in other parts of Asia. 
 
How is AI influencing the growth and future of data centers specifically in India?

AI data centers differ from traditional enterprise cloud data centers in three key aspects: scale, power density, and location. 

Firstly, AI data centers require much larger scale compared to typical cloud centers. This scale can be understood in two ways: the physical space and power required for operations, and the ability to deliver infrastructure across multiple countries. Providers like PDG offer solutions that span across various markets, which is advantageous for global tech companies looking to deploy several gigawatts of capacity simultaneously in different regions. This global reach allows companies to rely on a single provider to meet their needs in multiple markets. 

Secondly, power density is a significant differentiator. AI workloads use Graphics Processing Units (GPUs), which are far more energy-intensive than traditional computing equipment. These GPUs are often placed close together, working in parallel to solve large problems. This clustering leads to concentrated power and heat pockets, requiring advanced cooling technologies to maintain operational efficiency. The need for specialised cooling makes AI data centers distinct from traditional ones, where power and heat management are less complex. 

Finally, location plays a role. AI data centers are generally more flexible regarding latency requirements than enterprise cloud data centers. While enterprise data centers are often located near city centers or on the outskirts to minimise latency, AI data centers can be situated further away, providing more flexibility in their placement without compromising performance. These differences reflect the unique needs of AI infrastructure compared to conventional enterprise cloud systems. 
 
PDG recently announced a $1 billion investment for AI-ready data centers in India. Can you elaborate on the specific goals for this investment? 

Our expansion strategy focuses on following our clients to wherever they expand. The main drivers of this growth are enterprise cloud and AI. The $1 billion investment we've announced is for our long-term commitment, which includes adding 100 megawatts in Mumbai and 72 megawatts in Chennai. In Mumbai, we are expanding from 50 to 150 megawatts, making it one of the largest operational campuses in India. The Chennai location also has significant scalability potential beyond the land we've already acquired, allowing us to cater to large-scale demands. 

Another advantage we offer is local expertise. In every country, we have highly experienced teams. This helps us scale quickly, especially when it comes to securing land, permits, and aligning designs. 

When it comes to AI-grade data centers, flexibility in cooling is crucial. Different chips from companies like AMD, Nvidia, and Intel, used by Original Equipment Manufacturers (OEMs) such as Dell and Cisco, require different cooling methods. Some can be air-cooled, some need immersion in dielectric liquid, while others need heat exchangers. We deliver flexible designs, allowing for 80% standard construction and the remaining 20% tailored to specific cooling needs. 

In live data centers, we can also shift from one cooling method to another as needed. This is important because clients often experiment with different chips and cooling methods, and their requirements change over time. These flexible designs and fast scaling are key for AI infrastructure, all while ensuring sustainability and meeting the highest safety standards. 
 
With the AI revolution accelerating, how are PDG’s AI-ready data centers tailored to meet the growing computational demands of AI workloads, and what future-proofing measures are you implementing to adapt to evolving client needs? 

Our approach revolves around two key pillars: offering scale in a predictable way and flexible design. 

For example, in our Mumbai project, it took 14 and a half months from breaking ground to achieving Tier 3 certification, marking the completion of the entire data center facility. This shows how our local teams consistently deliver projects on time across regions. This reliability is what we offer clients looking for fast growth with flexibility. 

In terms of flexibility, we’re ready to support AI workloads. We can handle varying heat loads, with equipment running at 45 kilowatts per rack and designs capable of handling over 100 kilowatts per rack. We use different cooling methods, including conventional air cooling, rear cooling at the data center rack, and Direct-to-Chip Liquid Cooling (DCLC), where the cooling liquid reaches the tip of the rack. 

We stay ahead of the curve by ensuring our facilities are compatible with emerging technologies. For example, we recently piloted a two-phase immersion liquid cooling system, where the servers are submerged in liquid that vaporises into gas and then condenses back into liquid. 

We also participate in forums like the Open Compute Project (OCP), where hyperscalers share future requirements. PDG’s MU1 data center is the only OCP-certified facility in India, and our Singapore center hosts the only OCP lab in Asia. 

Finally, we maintain close relationships with hyperscalers, giving us early access to their technology plans. This ensures we involve them early in the design process, so when our data centers go live, we can deliver the cooling, flexibility, and service they expect. 
 
With hyperscalers like AWS and Google expanding their presence in India, how does PDG differentiate itself in providing hyperscale data center solutions, particularly regarding energy efficiency and scalability? 

When companies like Amazon, Google, and Microsoft set up data centers, they have two options. They can either use facilities like ours at PDG (Princeton Digital Group) to place their IT equipment, or they can build and operate their own data centers. Many of these companies have started purchasing land and constructing their own facilities. 

Globally, there's a balance between companies owning and operating their data centers and using third-party data centers, which is what PDG offers through co-location services. These two models will continue to coexist. 

What sets PDG apart is our "build to conviction" approach. We select locations and design data centers based on what we believe will meet the needs of companies like Amazon, Google, or Microsoft. This flexibility allows us to build facilities that can adapt to their specific requirements. As we begin or continue construction, these companies often need capacity quickly and turn to us for it. 

PDG doesn't compete with them in managed hosting or cloud services. Instead, we focus on co-location, offering precision cooling, uninterrupted power, security, and network services. 
 
Sustainability is a growing priority across industries. How is PDG balancing the demands for high-performance AI infrastructure with the need for energy-efficient, green data centers in India and beyond? 

Recently, PDG released its Environmental, Social, and Governance (ESG) report, which we have been doing for the past three years. PDG has committed to becoming net neutral for scope one and scope two emissions by 2030. Additionally, we have been tracking and reporting our carbon footprint over these years, and this year we started including scope three emissions as well. The report follows the Judicial Reinvestment Initiative (JRI) standards and aligns with the international Sustainable Development Goals (SDG) framework for our sustainability efforts. 

The JRI emphasises on the efficient use of data, monitoring of key metrics, and transparency in data reporting and analysis.

PDG has created a framework based on four pillars to enhance sustainability at every stage of the data center lifecycle. The first pillar focuses on maximising the use of renewable energy in all markets. Currently, our data centers operate on 50% renewable energy, with plans to increase this percentage. 

The second pillar is efficiency. We carefully select equipment that minimises power loss to improve overall energy efficiency. The third pillar is the choice of technology. For example, we prioritise air-cooled chillers over water-cooled ones in India, where water is scarce. 

The final pillar is certifications. In every market, we aim for the highest sustainability certifications. Our MU1 data center in Mumbai, for instance, is the first Indian Green Building Council (IGBC) platinum data center in the city, which is a significant achievement in sustainability. This framework helps guide PDG towards achieving our 2030 goal of net neutrality for scope one and scope two emissions. 
 
How does PDG leverage emerging technologies, such as edge computing or AI-driven infrastructure management, to enhance operational efficiency and customer service in its data centers? 

Currently, our main focus is on meeting the scalable requirements of hyperscalers. Edge computing will become more relevant as technologies like gaming, 5G, and autonomous vehicles gain traction. We're closely monitoring this space, and when the demand for edge computing grows significantly, we’ll make sure to leverage that market. For now, we are concentrated on large-scale, hyperscale-grade data centers in major cities. 

Regarding AI, before its adoption, our data centers already had sophisticated building management systems (BMS) in place. These systems capture real-time data on various parameters like temperature, humidity, power levels, and more. Some analytics are already built into these systems, helping us make informed decisions, such as optimising temperature settings and enhancing sustainability. The next step for us is to integrate and analyse this data across different centers using AI, which is currently in development. However, our existing systems already provide a solid foundation for efficient data center management. 
 
What role does India play in your overall growth strategy for Asia, and how do you see its role evolving as a digital hub over the next five to ten years? 

The process of India becoming a key digital hub for global needs has already started. Mumbai is emerging as a major location on the global map for AI players and hyperscalers planning their infrastructure strategies. This growth in Mumbai is driven by several factors, including the availability of land and power, and the strong support from local authorities. These conditions make Mumbai an attractive destination for AI infrastructure, similar to a few other locations across Asia. 

India's development as a regional digital hub depends on how data centers evolve in other countries and how much capacity India can continue to accommodate. Currently, India's data center market is around 900 megawatts, but ongoing projects are expected to more than double that capacity. The growth rate is already very fast. 

Data localisation and privacy regulations are becoming stricter in India and globally. How is PDG ensuring compliance with these regulations while maintaining flexibility for clients? 

We benefit from regulations that require data to be stored in India because it increases the demand for infrastructure, which we provide. When we talk about our role, we operate at the infrastructure level of data centers and don’t have access to the actual data or information stored on the servers. That belongs to our clients, and data privacy regulations apply to them. 

Our responsibility regarding data privacy is limited to managing the information of visitors, such as data entered into our registers or biometric details when someone enters our facilities. We've had strong processes in place for this for a long time to ensure compliance. However, the broader data protection laws you often hear about are not directly relevant to us. 
 
What key technology and market trends do you see shaping the future of data centers in India, and how is your company preparing to stay ahead? 

We are seeing a mix of influences from OEMs, chipmakers, enterprises, and government regulations. One expectation is increased support from the administration to boost renewable energy use. Each state has different regulations regarding renewable energy, which can create challenges in accessing it beyond a certain point. 

From a sustainability standpoint, we anticipate changes in this area. In terms of technology, various cooling systems are being tested. For example, we are looking at options like direct-to-chip liquid cooling and immersion liquid cooling. We believe one of these technologies will eventually become the standard, but currently, there is no global standardisation for immersion liquid cooling. Different vendors are promoting their own proprietary solutions. 

As data centers scale up, we need to ensure adequate infrastructure for power supply. For instance, a data center consuming 300 to 500 megawatts could require 600 to 800 megawatts due to overall consumption. This demand will affect both the state and national power grids. So, we see three key areas for evolution: increased use of green power, stabilisation of cooling technologies, and the necessary infrastructure support.