THE 15-STOREYED AI data centre of Yotta in Panvel, Navi Mumbai, hums round-the-clock, with the heat of artificial intelligence (AI). Inside the towering facility, the constant whirr of GPU processors from NVIDIA and CPUs from AMD and Intel fills the air. Pipes running vertically across the structure carry chilled water to cool the narrow server halls, where racks packed with processing units stand alongside storage systems from Dell, HP, and Asus.

On the rooftop, 34 industrial chillers — each with a capacity of 400 tonnes — operate continuously to maintain a temperature of 20-24°C inside the halls. Alongside the closed-loop chilling system, the 10-ft-tall server halls are designed with specialised air passages between aisles to regulate airflow and keep temperatures low.

Of the building’s 16 floors, only eight currently house the servers. Every alternative room is dedicated to power converters, motors, and water pipelines that sustain the complex cooling and electrical infrastructure. The tower has been powered by dual 110 kV (kilovolt) power stations, with nearly 70% of the electricity sourced from green energy.

The tower, named NM-1, runs on a power capacity of 75 MW and an IT load of 54 MW with odd and even architecture. Every odd floor is dedicated to utility (UPS and batteries) and even floor is dedicated to data racks, power distribution units (PDUs) and precision air handling units (PAHUs). The tower has been powered by dual 110 kV (kilovolt) power substations, with nearly 80% of the electricity sourced from green energy. As a backup, the centre relies on diesel generators — Yotta stores about 600,000 litres of diesel in underground tanks near the tower — and is connected to telecom networks through four redundant fibre paths, ensuring seamless connectivity.

The boom is visible. Yotta is constructing the second data centre with 80 MW capacity in the same campus. Spread across 100 acres in Navi Mumbai, it has the potential to accommodate 20-25 data centre buildings, with the capacity to scale up to 2 GW.

Yotta, started by the Hiranandani Group in 2019, is one of the earliest players to spot the business opportunity in the space. “What we saw early on was the convergence of three powerful forces — India’s rapid digitalisation, the growing need for sovereign data infrastructure, and the emergence of AI and cloud computing as foundational technologies for the global economy,” says Darshan Hiranandani, founder, Yotta Data Centre.

Within a short span, the company proved its business economics. “Our revenue has grown from $22 million in FY23 to $95 million in FY25. We’re projecting $166 million in FY26 and nearly $465 million in FY27, which represents a 118% CAGR from FY23 to FY27,” says Sunil Gupta, co-founder, CEO and MD, Yotta Data Services. More importantly, the profitability is scaling even faster. Ebitda improved from a loss of $9.8 million in FY23 to $31 million in FY25. “We expect Ebitda to grow to $119.9 million in FY26 and $365.2 million in FY27, reflecting a projected 147% CAGR.”

Around the time Yotta was betting big on data centres, another billionaire saw the same opportunity taking shape. Sunil Mittal-led Bharti Airtel carved out its data centre arm into Nxtra Data in 2020 and brought in The Carlyle Group as an investor, selling a 24.04% stake for $235 million the following year. Nxtra Data recently announced $1 billion fund-raising from Alpha Wave Global, Carlyle Global, Anchorage Capital, as well as its parent, in a deal that values the data centre firm at about $3.1 billion.

In FY25, Nxtra reported ₹2,079 crore in revenues and ₹224 crore in PAT. The company also recorded ₹772 crore in operating cash flows, while liabilities stood at ₹957 crore.

The real story, however, is just beginning to unfold. Corporate heavyweights — the Tata group, Reliance Industries Ltd, the Adani Group and Larsen & Toubro (L&T) — are now gearing up to ride the next wave of growth in data processing. A compilation of announced capital expenditure by these major groups, along with established players like CtrlS and Sify Infinit Spaces, indicates that India is set to invest $70 billion in building data centre infrastructure.

Sovereign cloud infrastructure and AI-ready data centres are critical components of the AI value chain, K. Krithivasan, CEO and MD, Tata Consultancy Services (TCS), tells Fortune India. “Our move to expand into the data centre business is a deliberate extension of our ambition to become the world’s largest AI-led technology services company. This will allow us to offer end-to-end AI infrastructure solutions for hyperscalers, Indian enterprises, and the government.”

Krithivasan feels there is a demand-supply mismatch for AI data centres in India. “Understanding and addressing this market gap creates an adjacency for TCS, which we believe will give us long-term returns,” he says.

The rush for data centres has a clear rationale. The Centre wants sensitive domestic data to remain within national borders instead of being stored in distant locations. Currently, Google, Microsoft, Meta, Amazon, and Oracle store a significant portion of their India-originated data in facilities located in the U.S., Europe, and Singapore.

India currently has over 150 data centres with about 1.5 GW operational capacity as of early 2026, including emerging AI-focussed facilities driven by GPU deployments, says K.G. Purushothaman, partner and head of technology transformation and AI, KPMG India. “The capacity to surpass 1.8 GW by the end of 2026 and reach over 8 GW by 2030 is fuelled by a shift toward AI-ready infrastructure encompassing hyperscale, specialised hardware like GPUs, software platforms, and edge computing.”

Budget 2026 strengthened the boom by offering a 21-year tax holiday — until 2047 — for foreign cloud companies that use Indian data centres and local resellers. The government’s broader goal is to create four-five large data centre clusters with a combined capacity of 5 GW by 2030. It is also working towards establishing eight data centre parks with a combined capacity of around 900 MW.

Yet India remains a relatively small player in the global data centre business. Worldwide, the industry is estimated to be worth around $348 billion in 2026, with North America accounting for nearly 39% of the market, driven by hyperscale cloud deployments and a powerful technology ecosystem. India’s market, in comparison, is still modest — valued at roughly $10 billion, with installed capacity of around 1.5 GW — and offers scope of expansion, as the country hosts nearly 20% of the world’s data, but only 3% of its storage.

Race of the giants

Three men who rarely let a new business opportunity slip by in India are Mukesh Ambani, Gautam Adani and N. Chandrasekaran. They see the AI data centre business as a natural extension of their existing strengths — construction, power generation, water management and cooling systems. Their groups also possess strong engineering capabilities, global partnerships with manufacturers of IT hardware, servers and semiconductors, and links to global hyperscalers. Added to this is connectivity to the world through undersea cable networks.

Alongside these heavyweights are sharp engineering and construction minds such as L&T chairman S.N. Subrahmanyan, who sees the data centre opportunity fitting perfectly into the group’s diverse portfolio. For companies with such capabilities, setting up a 200-MW AI data centre can be accomplished in less than a year.

Supporting this shift is the rapid evolution of chip technology. Semiconductor firms such as NVIDIA are introducing advanced GPUs built for parallel processing, enabling thousands of small cores to execute multiple tasks simultaneously. This provides massive computing throughput for data-intensive applications such as AI, 3D rendering, and complex simulations. Unlike traditional CPUs that process instructions sequentially, GPUs excel at SIMD (Single Instruction, Multiple Data) operations, breaking large problems into threads that run concurrently.

Prashant Chiranjive Jain, head, corporate centre at L&T, believes the opportunity is enormous. “The data centre market is likely to go upwards of 10-15 GW by 2030, and each GW entails an investment of $7-10 billion for conventional data centres and $15-35 billion for AI GPU-heavy data centres,” he says.

Within an AI data centre project, 25-35% of the cost typically goes towards EPC civil work and infrastructure, including land acquisition and regulatory permits. A much larger share — 60–70% — is spent on GPUs and IT hardware, while software and integration account for roughly 3–5%.

Ambani announced a $110-billion investment in the AI ecosystem, including the construction of a massive data centre of up to 3 GW near Reliance’s twin petroleum refineries in Jamnagar. RIL is also part of a three-way joint venture with Brookfield and Digital Realty to invest in data centres, targeting GW-scale infrastructure.

RIL’s AI data centres will be part of the newly launched Reliance Intelligence Ltd. “There is work going on on our own data centres that we need for captive purposes as well as for our partners in Jamnagar. We are also working towards gigawatt-scale data centres,” says Anshuman Thakur, head of strategy, Reliance Jio Infocomm. RIL has given a proposal to the Andhra Pradesh government to build a 1.5 GW data centre cluster along with a captive solar and battery storage system in Visakhapatnam. The overall cost could go upto ₹1.5 lakh crore.

Adani, too, has pledged an ambitious $100-billion investment by 2035 to build renewable energy-powered, AI-ready data centres across India, targeting a capacity of 5 GW. The initiative, expected to catalyse an additional $150 billion in spending across server manufacturing, sovereign cloud platforms, and supporting industries, focusses on sustainable, high-density computing. It aims to create a $250-billion AI infrastructure ecosystem in the country, integrating 30 GW of renewable power along with advanced storage solutions.

“The world is entering an intelligence revolution more profound than any previous industrial revolution,” Gautam Adani said in a statement earlier.

L&T, meanwhile, aims to capture up to 10% of the market over the long term through its entry into the AI factory space. The group’s engineering and construction expertise is supported by its IT services arm, LTM (earlier LTI Mindtree), as well as L&T Technology Services. Another group company, L&T Realty, can complement the venture through land development and real estate capabilities.

For the Tata group as well, the data centre business represents a powerful opportunity to leverage internal synergies. The combined strengths of Tata Electronics, Tata Power, Tata Realty, Tata Communications, Tata Tele and Tejas Networks can be utilised for the business. TCS is already building high-density, liquid-cooled data centres in India, with an initial capacity of 100 MW and plans to expand to 2 GW, in partnership with the U.S.-based alternative asset management firm TPG and semiconductor major AMD.

TCS has planned a phased, multi-year investment to build data centres. “The first 100-200 MW phase will carry a clear capex envelope, and the broader GW-scale programme is planned over the next five-seven years. HyperVault (the subsidiary) will be funded through a mix of equity and debt. We have brought in TPG as an equity partner,” says Krithivasan.

Together, TCS and TPG will invest up to ₹18,000 crore ($2 billion) over the next few years. Of the total commitment, TPG will invest up to ₹8,820 crore ($1 billion), and is likely to have a shareholding between 27.5% and 49% in HyperVault.

‘’As these capabilities mature, we’ll export that expertise to other markets. Like with all our investments, you will see the same discipline you have come to expect from TCS over the years: measured investments, prudent pacing, and a focus on durable shareholder value,” Krithivasan adds.

Despite the excitement, the economics of the sector remain complex. According to Jain of L&T, margin spreads can vary widely depending on the model—particularly for GPU-as-a-service offerings. They may range from 15-20% in co-location models to as much as 60-70% in specialised services, largely because of the heavy capital expenditure and rapid depreciation of hardware.

Bharti Airtel’s Nxtra operates 14 core facilities and about 120 EDGE data centres across 65 Indian cities, with a 12% share of the domestic market. Nxtra aims to increase it to 25% in the next three to four years, according to group executive vice chairman Gopal Vittal.

The surge in demand is also driving Chennai-based Sify Infinit Spaces, which operates 14 data centre facilities with a capacity of 188 MW. “We are developing 11 new data centres in strategic locations to address India’s growing demand,” says CEO Sharad Agarwal.

Hiranandani says the opportunity ahead is even larger. With the government announcing tax incentives for global data centre customers through 2047, and with India offering strong fundamentals such as reliable power, extensive fibre connectivity, and a robust safety and regulatory framework, the country is emerging as a global hub for digital infrastructure, he adds.

Rush of hyperscalers

Global tech giants are also pouring in tens of billions into building the country’s digital future.

During his last visit to New Delhi in December, Microsoft CEO Satya Nadella announced a $17.5-billion investment in AI infrastructure, while Amazon made a rival investment offer of $35 billion in AI projects across India. Prior to that, Google had committed $15 billion, in partnerships with the Adani Group and Bharti Airtel. Meta is also in talks with corporate houses to set up data centre facilities.

Microsoft is looking to expand cloud and AI infrastructure, for creating the company’s largest hyperscale footprint anywhere in the world. Amazon, the U.S. e-commerce giant, is accelerating efforts to build a comprehensive AI ecosystem to support business expansion across its India operations, with a focus on AI-driven digitisation, export growth and job creation. Of the $35 billion, building local cloud and AI infrastructure is expected to cost $12.7 billion.

“By integrating AI across consumer platforms, logistics networks and seller ecosystems, Amazon aims to reshape how digital commerce operates in one of the fastest-growing markets,” says Rajeev Rastogi, vice president, machine learning, Amazon India.

Google, meanwhile, is building a massive AI hub in Visakhapatnam with an investment of around $15 billion over five years. The facility will include GW-scale data centres supported by subsea cable networks and renewable energy, CEO Sundar Pichai had said during the project announcement. Described as the company’s largest investment outside the U.S., the project will focus on full-stack AI infrastructure, cloud computing, and a new global subsea cable gateway.

Another boom in sight?

For India, the stakes extend far beyond the data centre industry itself. The country hopes to replicate in AI the success it achieved in the global IT services sector three decades ago. Instead of writing software for overseas companies, India now wants to host and power the computing infrastructure that runs AI.

The Navi Mumbai region alone hosts around 64 data centres operated by 15 providers. Another major hub is Chennai. Together, the Mumbai and Chennai regions account for more than 70% of India’s data centre capacity.

Globally, the data centre market is projected to reach $802 billion by 2033, expanding at a CAGR of 12.7% as demand for cloud computing, AI workloads, and digital infrastructure accelerates. The Asia-Pacific region is expected to grow even faster, at an 18.5% CAGR, driven by rapid digital transformation, widespread 5G adoption, and increasing cloud usage across emerging economies. Hardware will dominate the data centre ecosystem, capturing more than 56% of the component market with a value exceeding $195 billion. The surge is being driven by massive investments in servers, GPUs, storage systems and networking equipment, according to a study by Persistent Market Research.

Meanwhile, software platforms used to monitor and manage facilities — known as data centre infrastructure management (DCIM) systems — are also witnessing rapid growth. Providers such as Schneider Electric (EcoStruxure), Vertiv (Environet/Avocent), Sunbird Software, and Nlyte (CommScope) are expected to see annual growth of around 16.7% as operators deploy advanced monitoring, automation, and energy management tools.

Three key types of data centres are gaining prominence — CPU-led centres for traditional workloads, GPU-led facilities optimised for AI training and inference (with high-density racks up to 30 KW), and edge data centres enabling low-latency services for sectors like finance, media, and 5G applications in Tier-II cities and towns. “Standard data centres currently yield Ebitda margins of 20-30%, while AI-optimised centres, with premium pricing for high-performance compute, are expected to achieve margins in excess of 50% due to elevated demand and utilisation,” says Purushothaman of KPMG India.

India’s edge data centre market is growing fast, enabling low-latency applications like 5G, IoT, and AI by bringing compute closer to users. The current capacity of edge centres is around 60-100 MW (around 5% of the total), which is expected to triple to over 200 MW by 2027, says Srinivas Varadarajan, founder and CEO, Vigyanlabs, a part of the micro data centres. “The market, valued around $600-900 million, could grow to $3-10 billion by early 2030s. Growth is driven by 5G rollout, data localisation, and rising digital demand in Tier-2 and 3 cities,” he adds.

The shift towards hyperscale and co-location facilities is transforming how companies build and operate digital infrastructure. Instead of running their own server rooms, enterprises are increasingly renting computing capacity from specialised operators. At the same time, rising energy consumption is pushing the industry towards greener solutions — from advanced liquid cooling systems to renewable power integration and more sustainable data centre architectures.

India already produces large numbers of AI engineers, data scientists, and software developers who work across global technology companies. What was missing was access to large-scale computing infrastructure within the country. Training advanced AI models requires enormous clusters of GPUs, high-performance networking, and vast storage capacity. Much of this infrastructure historically existed outside India, limiting domestic research and innovation.

The situation is beginning to change as both government and private sector investments accelerate.

The economic multiplier could be substantial. Data centres attract construction projects, networking infrastructure, equipment manufacturing, and specialised engineering services. They also support entire digital ecosystems — from cloud startups to artificial intelligence research labs. The industry’s rapid expansion could create thousands of high-skill technology jobs.

Yet success will depend on careful planning. Data centres require land, power, and water — resources already under pressure. Sustainable engineering, renewable energy integration, and efficient cooling technologies will become essential as the industry expands.

If managed well, however, the opportunity is enormous. India is entering a new phase of technological infrastructure building — one where invisible factories filled with servers may become as important as steel plants or refineries once were.

Inside these buildings, algorithms are trained, data analysed and digital services powered — systems that increasingly shape everyday life. In those blinking lights and humming machines lies the foundation of India’s next technological ambition: to ride the AI revolution and perhaps create another boom like the one that transformed the country’s IT industry a generation ago.

Challenges ahead

The rapid expansion of AI infrastructure also raises questions — particularly about the enormous physical resources required to run these digital factories.

AI computing is extremely energy intensive. Producing just 100 words of AI-generated text can consume roughly 0.14 kilowatt hours of electricity — enough to power around 14 LED bulbs for an hour. Cooling the computing equipment may require about half a litre of water.

With billions of users interacting with AI systems every day, the cumulative impact is significant. As of 2025, AI chatbots were receiving an estimated 2.5 billion prompts daily. The global AI chatbot market has grown to $11 billion, with nearly a billion users worldwide.

Behind these services lie thousands of data centres operating continuously. The expansion is beginning to face resistance in several parts of the world. Across the U.S. and Europe, local communities have increasingly opposed data centre projects because of their enormous electricity consumption and water usage. Residents worry about the strain on local power grids, environmental impacts, and the diversion of water resources in already stressed regions. Some cities have introduced stricter regulations on new facilities, while environmental groups have demanded tighter limits on energy and water use. These concerns are likely to intensify as artificial intelligence workloads expand.

Data centre business models depend on extremely high-density infrastructure — clusters of GPUs connected through high-speed networks and operating continuously to train and run AI models. Revenue typically comes from long-term contracts measured not by server racks, but the amount of power consumed, often leased in KW or MW. Moreover, the system prioritises uninterrupted electricity supply, ultra-fast networking, and extensive cooling infrastructure.

All these raise a critical question for India: can the country support large-scale domestic AI data processing while balancing energy, land, and water resources?

Initiatives such as the IndiaAI Mission and significant private sector investments in AI infrastructure are helping the country build sovereign computing capacity.

Beyond physical infrastructure, cybersecurity frameworks and compliance certifications are critical. India’s Digital Personal Data Protection Act (DPDP), along with sectoral regulations from the Reserve Bank of India (RBI) and the Securities and Exchange Board of India (Sebi), is strengthening compliance expectations and increasing confidence in domestic data processing and governance. By hosting infrastructure locally, companies can meet data-localisation requirements while maintaining control over security protocols.

Operators argue the industry is moving towards more sustainable designs. Modern hyperscale facilities use closed-loop cooling systems that recycle water and reduce consumption. Advanced energy management systems optimise power usage across servers and cooling infrastructure.

Renewable energy integration is also becoming central. Some data centre campuses already source a majority of their electricity from solar/wind power. Operators are exploring energy storage technologies and grid-balancing systems to improve efficiency and reduce carbon emissions.

At the same time, industry leaders emphasise the strategic importance of building digital infrastructure domestically. Data centres are increasingly viewed as critical national assets because they enable the entire digital economy.

The challenge will be to expand this infrastructure responsibly.