Sustainable Power Prediction and Demand for Hyperscale Datacenters in India ^†

Abstract

Data localization, data explosion, data security, data protection, and data acceleration are important driving forces in India’s datacenter revolution, which has raised a demand for datacenter expansion in the country. In addition, the pandemic has pushed the need for technology adoption, digitization across industries, and migration to cloud-based services across the globe. The launch of 5G services, digital payments, big data analytics, smartphone usage, digital data access, IoT services, and other technologies like AI (artificial intelligence), AR (augmented reality), ML (machine learning), 5G, VR (virtual reality), and Blockchain have been a strong driving force for datacenter investments in India. However, the rapid expansion of these datacenters presents unique challenges, particularly in predicting and managing their power requirements. This abstract focuses on understanding the power prediction and demand aspects specific to hyperscale datacenters in India. The study aims to analyze historical power consumption data from existing hyperscale datacenters in India and develop predictive models to estimate future power requirements. Factors such as server density, workload patterns, cooling systems, and energy-efficient technologies will be considered in the analysis. Datacenter negatively impacts the environment because of the large consumption of power sources and 2% of the global contribution of greenhouse gas emissions. Given the increasing cost of power, datacenter players are naturally encouraged to save energy, as power is a high datacenter operational expenditure cost. Additionally, this research will explore the impact of renewable energy integration, backup power solutions, and demand–response mechanisms to optimize energy usage and reduce reliance on conventional power sources. Many datacenter providers globally have started using power from renewable energy like solar and wind energy through Power Purchase Agreements (PPA) to reduce these carbon footprints and work towards a sustainable environment. In addition, today’s datacenter industry constantly looks for ways to become more energy-efficient through real innovation to reduce its carbon footprint.

1. Introduction

In the current digital transformation era, the heavy emphasis on data has grown like never before, from day-to-day business activities to financial transactions. Datacenters have evolved as economic warehouses in the digitally connected world as life has become increasingly data-driven [1]. They are not only storing the data, but data accessibility, data security, data processing, and data analytics are also a prime concern. Today, many businesses and government agencies rely on datacenters to securely store critical operational and proprietary assets in a centralized digital environment. Also, the epidemic has accelerated digital processes. This transition to the cloud has resulted in greater investments in hyperscale datacenters [1].

The growing number of internet users and mobile penetration, the government’s emphasis on e-governance/digital India, the adoption of new technologies (cloud computing, artificial intelligence (AI), the Internet of Things (IoT), machine learning (ML), 5G, augmented reality (AR), virtual reality (VR), and Blockchain, among others), and the growing user base for social media, gaming, e-commerce, and OTT platforms are the key triggers for the digital explosion in India [2]. This, together with favorable economic regulatory policies such as the draught Digital Data Protection Bill 2022, proposes to offer incentives worth up to Rs. 15,000 crores under a national policy framework for datacenters and provide infrastructure status to datacenters [3] as well as unique incentives from central and state governments, such as subsidized land, power subsidies, stamp duty exemptions, discounts on the use of renewable energy, and procurement of IT components made locally. The government plans to invest up to Rs. 3 lakh crore in the centerfire ecosystem over the next five years, according to a draught policy. As the budget 2022–2023 was announced, the finance minister granted the datacenter industry infrastructure status.

Many digital initiatives have been introduced by the Indian government to encourage the localization of data. Storage of Payment System Data (2016), Insolvency and Bankruptcy Code (IBC) (2016), and the National e-Governance Plan (NeGP) (2006) are all examples of such laws and regulations (2018). Increased demand for data storage is anticipated as a result of the completion of the Draft E-commerce Policy (2019), the Draft Data Protection Law (2019), and the proposed policy on DC parks (2020) [3,4].

The Government of India has also taken several digital initiatives that have encouraged the creation of a robust digital foundation, like the following.

• Digital India: The government’s attempts to modernize healthcare, education, banking, agriculture, citizen services, and smart cities, among other areas, are aimed at making India a trillion-dollar digital economy. All of this will need a strong underlying digital infrastructure [2].
• Enterprise digital transformation: As customer experiences and operational efficiency business excellence become more important in boardroom discussions, businesses will increasingly approach AI, IoT, AR, and VR to innovate and realign themselves with the preferences of the digital-native consumer. The core of these new-age use cases will be high-performance computing and the cloud, providing datacenter providers with a variety of chances to cater to [2].
• Digital banking and payments: India’s adoption of digital payments facilitated by the Unified Payments Interface (UPI) is a global case study. In 2021–2022, the country’s largest retail payment system completed transactions worth more than Rs. 76 trillion, up from Rs. 41 trillion in 2020–2021. According to Mr. Shaktikanta Das (Reserve Bank of India (RBI) Governor), the figure is likely to reach Rs. 100 trillion [2].
• E-commerce and OTT: COVID-induced lockdowns have played a significant role in attracting more individuals to e-commerce and OTT platforms, and this trend is likely to continue. By 2030, the online retail industry is predicted to account for 37% of the entire organized retail market, highlighting the magnitude of transactions that e-commerce platforms will see and how datacenters and the cloud will be at the forefront of enabling it [2].
• Data localization: Future data expansion will be largely driven by the decision to keep Indian data within the country’s borders, which would necessitate the construction of hyperscale datacenter clusters throughout India [2].

Information is crucial to businesses that want to remain competitive. This is causing a rise in the need for data storage, which in turn is fueling the worldwide expansion of the datacenter industry. Arizton predicts that the value of the Indian datacenter industry will increase from its 2021 valuation of $4.35 billion to $10.09 billion by the year 2027 at a compound annual growth rate (CAGR) of 15.07%. There were 138 datacenters in India as of March 2022, and that number is expected to grow to 183 by 2025. Despite steady expansion over the past few years, the United States continues to lead the world with more than 2700 datacenters. It is followed by Germany, which has 487 datacenters. In terms of the total number of datacenters, India is ranked 13th worldwide. Datacenters are notorious power consumers, and in India, the datacenter industry is predicted to increase from its current 637 MW in H1 2022 to 1318 MW by 2024. The ICRA report predicts that over the next six years, an additional 4900–5000 Megawatts of datacenter capacity will be installed at an estimated cost of roughly Rs 1.5 trillion [2]. Compared to the existing levels, this is a six-fold rise. Thus, we can estimate the necessary power capacity to meet the datacenter’s uninterruptible, multi-source power demand.

2. Datacenter and Type of Datacenter

A datacenter is a facility housing a network of servers, storage devices, and other computing equipment and is used by organizations for centralized data storage, management, processing, and distribution. Powerful technologies are used in today’s datacenters to handle an organization’s data, as illustrated in Figure 1. The HVAC system, power distribution units (PDUs), uninterruptible power supply (UPS), backup generators, fire suppression systems, and security systems are all examples of non-IT components that are essential to the operation of datacenters alongside IT components [5].

To ensure 99.995 percent uptime, scalability, and reliability for today’s mission-critical applications and data, the information and communications industry and datacenters are adopting the obligations imposed by the ANSI/TIA-942 and the Uptime Institute Tier Topology. The Uptime Institute Tier Topology covers only mechanical and electrical systems, whereas the TIA-942 is essential for modern mission-critical datacenters because of its comprehensive coverage of all datacenter facilities (site location, architectural, mechanical, electrical, telecommunication, physical security, fire safety, etc.). The datacenter is subject to audits and evaluations by these institutions according to specific standards, and its availability is certified accordingly. The Tier classification system offers a standardized approach for comparing diverse and personalized datacenter facilities based on the expected performance and uptime of their site infrastructure, details, and classification, as shown in

Table 1. Tia-942 Datacenter Tier Classification System.

Fields	Tier I Basic	Tier II Redundant Components	Tier III Concurrently Maintainable	Tier IV Fault Tolerant
Number of Delivery Paths	Only 1	Only 1	1 Active 1 Passive	2 Active
Redundant Components	N	N + 1	N + 1	2(N + 1)S + S
Support Space to Raised Floor Ratio	20%	30%	80–90%	100%
Initial Watts/Ft	20–30	40–50	40–60	50–80
Ultimate Watts/Ft	29–30	40–50	100–150	150+
Raised Floor Height (Inch)	12′′	18′′	30–36′′	30–36′′
Floor Loading Pounds/Ft	85	100	150	150+
Utility Voltage	208,480 V	208,480 V	12–15 KV	12–15 KV
Months to Implement	3	3 to 6	15 to 20	15 to 20
Year First Deployed	1965	1970	1985	1995
Construction Rupees/Ft Raised Floor	₹40,000	₹40,000	₹40,000	₹40,000
Annual IT Downtime due to failure	28.8 h	22.0 h	1.6 h	0.4 h
Site Availability	99.671%	99.749%	99.982%	99.995%

.

2.1. Type of Datacenters

Different datacenter and datacenter service model options are available, which are classified based on whether they are controlled by a single company or a group of companies. They are also classified based on who manages the datacenter core components like compute, network devices and storage technologies, energy efficiency system, and security systems.

2.1.1. Hyperscale Datacenters

Hyperscale datacenters are modular in nature and are built to ensure high efficiency and scalability. Modularity allows enterprises to take a step-by-step approach to building their datacenters. Hyperscale datacenters are also built for high redundancy—a critical ability for datacenters. Hyperscale datacenters are much larger than enterprise datacenters; a hyperscale datacenter should have more than 6000 servers and 10,000 square feet of space.

2.1.2. Cloud Datacenters

The cloud datacenter is rapidly becoming the preferred mode of data storage for medium and large enterprises. This is because it is much more secure than using traditional hardware devices to store information. Cloud datacenters provide a high degree of security protection, such as HSM, unified threat management firewalls, Antivirus, DLP, PPM, and backup components, in the event of a security breach. A critical part of moving to the cloud is to prioritize which applications or workloads have to move to the cloud. This depends on factors such as performance, security, availability, and compliance.

2.1.3. Enterprise Datacenters

Enterprise datacenters are constructed, owned, and operated by organizations and optimized for their end-users, typically on the corporate campus or offsite. Enterprise datacenters are essential for organizations that rely heavily on technology to support their business operations and require a reliable and secure infrastructure to manage their critical data and applications.

2.1.4. Edge Datacenters

Edge datacenters are decentralized facilities that provide cloud services and resources to internet users. They improve user experience by delivering cached content faster and at a lower cost. By 2025, 70% of organizations will focus on small data stored in edge computing devices, enabling faster processing for low-latency applications. Hosting servers closer to end users can reduce latency and improve access speed, leading to higher customer conversion rates.

2.1.5. Colocation Datacenters

A colocation data center allows businesses to rent infrastructure, power, cooling, security, and bandwidth for their servers. This is especially useful for businesses that do not have their own data center. Retail and banking sectors are increasingly turning to colocation data centers so that they can concentrate on their core business instead of managing the data center.

2.2. Important Components to Build the Hyperscale Datacenter

The most important factors in choosing a site for a datacenter are its geographic position, favorable climate, the availability of power, its accessibility to clients, fibers connection, and the cost of the property.

• Scale and Power: 62,000 m² of gross space, allowing you to scale up or down as your needs evolve, complemented by 120 MW available to power your IT infrastructure.
• Connectivity: 100% carrier, cloud, and internet exchange neutral to benefit from better cost efficiencies, improved redundancy, and optimal uptime, with multi-cloud access available on demand.
• Location: Perfectly situated to access other highly connected datacenters, multiple diverse routes to key carrier hotels, and a high concentration of fiber routes.
• Excellence: Our standards are not just tied to our technical capabilities but extend to our people, which means we have the best resources to help you achieve your IT infrastructure goals.
• Sustainability: A significant proportion of renewable energy to support our customer’s sustainability goals alongside our own.
• Market Access: To the Indian market with the right resources and experienced staff already in place, jurisdiction laws and local market legislations will not be obstacles.

2.3. Datacenter Services in India

The largest Tier IV datacenter in Asia, certified by the Uptime Institute and situated in Navi Mumbai, was opened by Yotta in July 2020. A month later, Web Werks opened its fourth datacenter in Pune. Whereas ESDS offers cloud services, smart electrical meters, smart city solutions, and various AI services, NxtGen offers transformative services, including DevCloud, machine learning (ML), and artificial intelligence (AI). It is important to note that Indian datacenters use less energy than those located in other parts of the world. Additionally, several datacenters, including Yotta and CtrlS, have their own alternative energy sources, including gas turbines and solar farms. As of February 2021, Savills research shows in Figure 2 that the megawattage of datacenters per consumer in India is fifteen times lower than in Europe. Indian datacenters only require 500 MW of power for their over 500 million customers, compared to 8600 MW in Europe for its 600 million users.

Since 2008, India has benefited from $977 million in private equity and other strategic investments in the datacenter sector. Of this, $396 million arrived in 2020 up until September [6,7,8]. The market for datacenters is predicted to be driven by an increase in data traffic, government-backed initiatives, and looming data sovereignty regulations. High volumes of data and the need to protect it from cyber threats have pushed Inputs to accelerate its engagement with global industry giants in order to accomplish the goal of developing the critical assets of the IT sector as soon as possible. India has a long way to go before it can be considered a global datacenter hub [1].

2.4. Top Ten Datacenter Operators in India

ESDS: ESDS (Nashik, India), founded in 2005, is one of India’s leading auto-scale patented cloud and managed service providers. ESDS has its own datacenters with a capacity of 2 MW in Nashik and Mumbai, as well as joint ventures with STPI in Bangalore and Mohali. ESDS is constantly growing its customer base by providing a variety of services, such as managed datacenter solutions, virtualization, disaster recovery hosting, and managed cloud solutions, all with technical support. Manufacturing, education, banking and finance, eCommerce, energy and utilities, healthcare, agriculture, IT, entertainment and media, telecom, government, and travel and tourism are just a few of the industries where the company has a presence [1].

CtrlS: CtrlS (Hyderbad, India), based in India, is one of Asia’s largest managed service providers and ANSI/ITA-942 Tier 4 certified datacenters. It operates seven cutting-edge facilities in Hyderabad, Mumbai, Noida, and Bangalore, totaling 1.2 million square feet and 124.3 MW of power. It serves 60 Fortune 500 global multinationals, according to the company. The CtrlS Mumbai datacenter facility is LEED Platinum certified v4 O+M by the United States Green Building Council (USGBC) and is powered by 1 MW of solar panels. Its datacenter in Noida is completely earthquake-proof and pollution-free. CtrlS intends to increase its footprint by five million square feet over the next few years, including the construction of a two million square foot, 250 MW, 24,000 rack capacity hyperscale datacenter in Navi Mumbai and Hyderabad, as well as a one million square foot, 120 MW hyperscale datacenter in Chennai [1].

GPX Global Systems Inc [Equinix]: GPX (Mumbai, India) develops and operates private, carrier-neutral Tier 4 datacenters in Mena and South Asia. GPX offers secure and highly reliable carrier-neutral datacenters to both domestic and international clients who need to consolidate their critical business infrastructure. Equinix intends to construct two facilities, MB1 and MB2, with an initial capacity of 1350 racks and an additional capacity of 500 racks. More than 90,000 square feet (8200 square meters) of colocation space will be available at the facilities. GPX Mumbai 2, a Tier 4 facility spanning 6000 m² with a total capacity of 16 MW, went live in the second quarter of 2019 and serves a variety of customers, including cloud service providers, Telcos, CDNs, Internet Service Providers, e-businesses, and enterprise clients.

Netmagic: Netmagic (Mumbai, India) is a leading provider of managed hosting and multi-cloud hybrid IT solutions, and it is completely owned by NTT Communications. Its nine carrier-agnostic, hyperscale, and high-density datacenters provide services to more than two thousand businesses throughout the world. In addition, business customers in the Americas, Europe, and Asia can make use of Netmagic’s Remote Infrastructure Management (RIM) services. Its main office is in the Indian city of Mumbai. Five of Netmagic’s datacenters are located in Mumbai, two in Bengaluru, and one each in Chennai and Noida has recently gone live, bringing the total number of datacenters it runs in India to nine. Almost 30 MW of IT demand will be supported by the NAV1A datacenter, which will occupy 400,000 square feet (37,161 square meters) (G+8 structure with one basement floor). Up to four datacenters with a combined 150 MW of power can be housed on the NAV1 campus in Mahape, Navi Mumbai [1].

NxtGen: Netmagic (Mumbai, India), a subsidiary of NTT Communications, is a prominent supplier of managed hosting and multi-cloud hybrid IT solutions. It includes nine carrier-neutral datacenters, including hyperscale and high-density facilities, and serves over 2000 organizations globally. Netmagic, located in Mumbai, also provides Remote Infrastructure Management (RIM) services to business customers worldwide, including NTT Communication customers in Europe, the Americas, and Asia-Pacific. In India, the NxtGen High-Density Datacenter TM (HDDC) is designed to fulfill Tier III criteria, hosting up to 2000 high-density racks in four separate datacenters with a per-rack capacity of 15 KW. NetGen intends to create enterprise and edge datacenters in the near future.

Nxtra Data Limited: Nxtra Data Ltd. (Delhi, India) was established to oversee Bharti’s datacenter-managed services operations. Nxtra currently oversees 10 datacenters in Manesar, Noida, Chennai, Mumbai, Bangalore, Bhubaneswar, and Pune, all of which have achieved Tier III status and ISO 27001 certification. The combined space of all the buildings is around 200 thousand square feet. Earlier this month, Bharti Airtel Group stated that it would be investing Rs 2000 crore in the region to build a massive hyperscale datacenter—hyperscale datacenter with an initial IT load capacity of 60 MW.

Yotta: Yotta (Mumbai, India), powered by the Hiranandani Group, designs, builds, and operates exponentially scalable datacenter parks. Yotta’s 50+ Acres of datacenter parks include 11 datacenter buildings with options ranging from a single rack to a complete building or even a customized DC, all supported by a variety of managed services. Yotta has recently built two large hyperscale datacenters in Noida and Mumbai Panvel, each having a capacity of 5000 racks, 300,000 square feet, 28.8 MW electricity, and 7200 racks with 50 MW power. Yotta Greater Noida, a datacenter park, will be expanded with a further six interconnected datacenter buildings featuring 30,000 racks and 160 MW of Power. Also, the Yotta NM1 datacenter features 7200 racks and 30.4 MW of power [2].

AdaniConneX: AdaniConneX (Siruseri, India), a joint venture of Adani Enterprise and EdgeConneX, has begun operations at its first hyperscale datacenter complex, Chennai 1, in Chennai’s SIPCOT IT Park. The campus will initially have a capacity of 17 MW (IT Load), which will be ramped up to 33 MW at full capacity. This is Tamil Nadu’s first IGBC Platinum Rated datacenter, and it will be fueled entirely by renewable energy [9,10].

2.5. Datacenter Power Requirement and Future Demand

Currently, coal, lignite, gas, and diesel are fossil fuel types of resources to generate power along with non-fossil fuel sources like hydro, wind, solar, BM power/cogen, waste to energy, and small hydropower. As shown in

Table 2. Installation power generation capacity report. Source: Ministry of Power.

CATEGORY	INSTALLED GENERATION CAPACITY (MW)	% of SHARE IN Total
Fossil Fuel
Coal	205,235	49.10%
Lignite	6620	1.6%
Gas	24,824	6.00%
Diesel	589	0.10%
Total Fossil Fuel	237,269	56.80%
Non-Fossil Fuel
RES (Incl. Hydro)	173,619	41.40%
Hydro	46,850	11.20%
Wind, Solar and Other RE	125,692	30.20%
Wind	42,868	10.30%
Solar	67,078	16.10%
BM Power/Cogen	10,248	2.50%
Waste to Energy	554	0.10%
Small Hydro Power	4944	1.20%
Nuclear	6780	1.60%
Total Non-Fossil Fuel	179,322	43.00%
Total Installed Capacity (Fossil + Non-Fossil)	417,668	100%

, the carbon mission of fossil fuel sources is more as compared to renewable energy sources. As per the current report, 57.4% of power is generated using fossil fuels, and 42.6% is generated using renewable energy.

According to the Ministry of Power, the Government of India is adhering to the power Need and Availability report shown in

Table 3. The power supply position in the country during 2009–10 to 2023–24.

	Energy			Peak
Year	Requirement	Availability	Surplus (+)/ Deficts (−)	Peak Demand	Peak Met	Surplus (+)/Deficts (−)
(MU)	(MU)	(MU)	(%)	(MW)	(MW)	(MW)	(%)
2009–2010	830,594	746,644	−83,950	−10.1	119,166	104,009	−15,157	−12.7
2010–2011	861,591	788,355	−73,236	−8.5	122,287	110,256	−12,031	−9.8
2011–2012	937,199	857,886	−79,313	−8.5	130,006	116,191	−13,815	−10.6
2012–2013	995,557	908,652	−86,905	−8.7	135,453	123,294	−12,159	−9
2013–2014	1,002,257	959,829	−42,428	−4.2	135,918	129,815	−6103	−4.5
2014–2015	1,068,923	1,030,785	−38,138	−3.6	148,166	141,160	−7006	−4.7
2015–2016	1,114,408	1,090,850	−23,558	−2.1	153,366	148,463	−4903	−3.2
2016–2017	1,142,929	1,135,334	−7595	−0.7	159,542	156,934	−2608	−1.6
2017–2018	1,213,326	1,204,697	−8629	−0.7	164,066	160,752	−3314	−2
2018–2019	1,274,595	1,267,526	−7070	−0.6	177,022	175,528	−1494	−0.8
2019–2020	1,291,010	1,284,444	−6566	−0.5	183,804	182,533	−1271	−0.7
2020–2021	1,275,534	1,270,663	−4871	−0.4	190,198	189,395	−802	−0.4
2021–2022	1,379,812	1,374,024	−5787	−0.4	203,014	200,539	−2475	−1.2
2022–2023	1,511,847	1,504,264	−7583	−0.5	215,888	207,231	−8657	−4
2023–2024	266,951	266,360	−591	−0.2	221,370	221,347	−23	−0.01

, which indicates that we are already at a minimum in terms of meeting current power demand.

2.6. Datacenter Power Usages Effectiveness

A typical datacenter power usage distribution is shown in Figure 3; majorly, IT equipment like computing, network, and storage devices and cooling systems need substantial energy. So, managing energy consumption in the IT load can directly affect the supporting systems, such as the power distribution units, cooling system, and UPS system, which ultimately enhance the Datacenters’ overall energy efficiency. The energy intensity of the IT load significantly affects the cooling system’s energy requirements. The use of high-processing systems such as AI–ML data processing, HPC systems, and big data analytics systems is increasing substantially, not only in the compute requirements but double cooling systems as well. Typically, the cooling system uses 35 to 40 percent of the datacenter’s electricity. As a result, behind the IT load, the cooling system consumes the second most energy.

The Green Grid’s widely adopted benchmarking standards for measuring datacenter power consumption performance, known as Power Usage Effectiveness (PUE) and DataCenter infrastructure Efficiency (DCiE), were designed to assist datacenter IT professionals in measuring the datacenter power effectiveness and take the initiative to improve the power consumption efficiency.

Power Usage Effectiveness is calculated by dividing the total amount of facility power entering a datacenter by the power consumed by the IT equipment.

$$\mathrm{PUE}=\frac{\mathrm{Total} \mathrm{facility} \mathrm{energy} \mathrm{usage}}{\mathrm{IT} \mathrm{equipment} \mathrm{energy} \mathrm{usage}}$$

(1)

A PUE of 1.0, which corresponds to 100% efficiency, indicates that all energy used is used exclusively for IT equipment without any distribution losses. However, it is extremely challenging to achieve such a level of efficiency. According to the most recent Uptime Institute report, the average PUE ratio for datacenters in 2020 was 1.58, which is only marginally better than the average ratio recorded seven years earlier.

DataCenter Infrastructure Efficiency (DCiE): It is a metric developed by the Green Grid for datacenter infrastructure efficiency. It is an efficiency measure that is calculated by dividing the IT device power consumption by the energy consumption of total datacenter usage, like the sum of all electrical energy used for IT devices, BMS systems, security systems, HVAC systems (steam or chilled water, PAC), a power distribution unit, lighting, etc. So, it is the inverse of the PUE calculation [5].

$$\mathrm{DCiE}=\frac{1}{\mathrm{PUE}} \times 100\%$$

(2)

(The power consumption of the IT devices divided by the power consumption of all datacenter equipment) × 100%.

2.7. Environment Footprint

To understand IT carbon footprints and make their IT operations green, business leaders must first understand what generates their carbon footprint. There are typically four main interrelated sources of carbon emissions: datacenters, big data and analytics, security and cryptography, and internet consumption [11].

According to the International Energy Agency, datacenters worldwide use 200 to 250 terawatt-hours (TWh) of electricity annually (IEA). The equivalent of 0.3% of all carbon emissions worldwide and around 1% of the world’s electrical demand. Future demand for datacenters and network services will only increase, resulting in more electricity use and carbon emissions. According to some estimations, datacenter operators’ need for electricity capacity increased by 43% in absolute terms between 2018 and 2021, and the global datacenter market will grow by more than 30% between 2021 and 2027. Therefore, action to support further efficiency improvements, lower energy consumption, and reduce carbon impact from datacenters is critical.

According to recent research by global real estate consultant JLL, which was cited in the EY report, datacenters now account for up to 4% of the world’s greenhouse gas emissions, and their energy consumption doubles every four years. The study also states that environmental, social, and governance (ESG) standards will have a direct impact on this market’s expansion [12]. Thus, improving our sustainability and social responsibility will be one of our major priorities during the next two years. Modern datacenters are using advanced technologies like AI and ML to address current challenges. When combined with automation and IoT, ML and AI can forecast datacenter power failures, maintenance costs, cooling requirements, power usages, power and cooling wastages, and improve performance metrics. A Gartner estimate claims that by 2025, half of all cloud datacenters will have implemented advanced robots with AI and ML capabilities, increasing operating efficiency by 30%. As Forges suggested, here are three ways to cut CO₂ commission and use a greener datacenter [12].

2.7.1. Use of Hydrogen Cell

Datacenters are looking to hydrogen gas turbines as a way to become green in communications hubs like Singapore. The steam methane reforming (SMR) process will be used to extract the hydrogen needed to run the turbines from fossil fuels. When burned, hydrogen emits no greenhouse gases, and it can be adapted to burn in natural gas power plants that can also burn natural gas.

2.7.2. Combined Heat and Power

Since heating and cooling are the two major energy consumers for datacenters, it is imperative to develop a strategy that is effective in addressing this need. The solution is offered by combining the power and cooling systems. MHI-AP and Keppel datacenters are looking into ways to lessen the impact of datacenters in Singapore on the environment by developing an energy-efficient “tri-generation” system that can provide heat, power, and cooling all at once.

2.7.3. Use of Renewable Sources

In the future, renewable energy will play a crucial role in the energy sector, but the major challenge is that renewable energy is available only for a particular period and require separate storage facilities to use 24 × 7; however, datacenters require uninterrupted “always on” power. Many businesses choose to offset for the time being in order to become carbon neutral [13]. Google datacenters are using storage facilities to store renewable energy, which is generated using solar parks and wind farms. They have developed AI- and ML-based algorithms to use more renewable electricity when it is available.

2.8. GreenIT Challenges

In the coming decade, significant government and industry initiatives are required to support energy efficiency, encourage the purchase of renewable energy sources, and promote green IT research, design, and development in order to enable datacenter growth that is both responsible and robust.

2.8.1. Complex Carbon Accounting

IT firms need tools for precisely measuring their carbon emissions if they want to lower the carbon footprints of their datacenters, apps, and infrastructure. Although they may not be widely available or fully functional at the moment, these tools are under development and will be crucial to the future greening of IT [14,15,16,17].

2.8.2. Opaque Benchmarks

Organizations need benchmarks, which are internal and external key performance indicators (KPIs), to provide them with a sense of their performance and growth, even if they have a clear and precise accounting of their carbon footprint. Clarity over acceptable performance measures and measurement and data collection standards will be necessary for this [14].

2.8.3. Ambiguous Analytics

Organizations will have a clearer understanding of where they are and where they want to go once they can easily, economically, and accurately monitor and benchmark the carbon footprint of their IT infrastructure. But, they still need to translate unclear analytics into firm action in order to go from point A to point B. To evaluate data from carbon accounting and determine what actions to take, where to take them, and when to take them, they need solutions and knowledge.

The topic of power prediction and demand for hyperscale data centers in India is of significant importance, considering the exponential growth of data storage and processing needs in the country. This study analyzes power consumption data for hyperscale data centers in India and develops predictive models for future power requirements. Collaboration between operators, policymakers, and energy providers is necessary for optimal power management. Implementing the study’s findings can make data centers more energy-efficient, reduce their carbon footprint, and contribute to sustainable development [18,19].

3. Conclusions

In recent years, technology adoption and digitization have rapidly increased in various industries in India, leading to a rise in datacenter investments and the number of datacenters. However, poorly planned datacenters can consume large amounts of energy, resulting in increased carbon emissions. Environmentalists have been advocating for the use of clean energy in datacenters, but optimizing power efficiency utilization is a complex task with many interdependent elements. Cloud service providers and datacenter players are expected to adopt affordable, cutting-edge cooling methods that use less power. By sharing accurate, complete, and timely data, ICT companies can help energy researchers and policymakers understand the impact of ICT service demand on energy use. However, there is a need for more transparent and robust tools for monitoring, measuring, and reporting power utilization, which can lead to more accurate results and further carbon emission reduction. Overall, the research on power prediction and demand for hyperscale datacenters in India highlights the significance of accurate power forecasting, efficient energy management, and sustainable practices in this vital sector. To reduce energy waste and improve efficiency in the datacenter, it is important to use tools like 360-degree AI–ML monitoring, real-time PUE ratio measurement, and advanced carbon measuring technology. These tools can help ensure effective energy utilization. Continued efforts in this area will pave the way for a robust and sustainable datacenter industry, ensuring India’s position as a leading player in the digital era.