India’s AI data center race meets reality

By mid-2026, Indian IT giants HCLTech and Tata Consultancy Services (TCS) embarked on a bold strategic pivot from their traditional asset-light software services to owning capital-intensive AI data center infrastructure, with HCLTech committing $364 million and TCS planning a massive $6-7 billion investment over the next decade. This shift is driven by the explosive rise of AI and India's stringent data localization laws, compelling these firms to reduce reliance on global cloud hyperscalers like AWS and Microsoft, and instead offer integrated AI services tailored to regulated industries through bundled infrastructure and consulting solutions.

This transition reflects a profound reimagining of the Indian IT business model, moving away from the traditional 'people, projects, and billable hours' approach toward a future centered on 'computing power, models, platforms, agents, and business outcomes.' However, the shift entails considerable risks: the capital-intensive nature of data center ownership threatens to strain balance sheets and compress profit margins, as noted by analysts from PhillipCapital and BOBCaps who caution that returns could fall into the low teens, far below TCS’s historical performance.

Moreover, the operational challenges of managing AI data centers are nontrivial, as expertise in AI software does not seamlessly translate into competencies required for hardware-intensive infrastructure such as power optimization, cooling, and supply chain management. Historical precedents like Microsoft’s costly mobile device venture and IBM’s exit from PCs underscore the pitfalls of vertical integration in hardware, highlighting that Indian IT firms must navigate a steep learning curve to successfully manage these complex data center operations.

In a landmark public-private partnership announced in July 2026, HCLTech joined forces with Sarvam AI and the Odisha government to launch a ₹14,257 crore Sovereign AI Park in Bhubaneswar, signaling a major stride toward India's AI sovereignty and data localization ambitions. This initiative not only includes the establishment of a cutting-edge AI data center but also the creation of a Global Technology Center expected to generate 5,000 jobs by 2028, positioning Odisha as a burgeoning technology and AI hub. The project’s formalization, attended by top leaders such as HCLTech Chairperson Roshni Nadar Malhotra and Odisha Chief Minister Mohan Charan Majhi, underscores its strategic significance in expanding India’s digital infrastructure.

The collaboration leverages HCLTech’s comprehensive AI expertise alongside Sarvam AI’s pioneering Indian-language foundation models to develop multilingual, sector-specific AI solutions tailored for both public and private sectors. This synergy exemplifies how combining technological capabilities with localized AI models can advance India’s goal of digital autonomy. Furthermore, Sarvam AI’s ambition to scale its GPU infrastructure from 2,000 to 10,000 Nvidia Blackwell GPUs, supported by government funding pursuits, highlights the dynamic public-private cooperation fueling sovereign AI infrastructure development under the IndiaAI Mission.

Larsen & Toubro-Vyoma has transformed from a traditional co-location provider into a sovereign AI infrastructure powerhouse, currently operating 32 MW of live IT capacity with ambitious plans to expand by over 220 MW. Their integrated technology stack—spanning data center infrastructure, sovereign cloud, AI-ready platforms, and managed services—caters to sectors such as BFSI, government, healthcare, manufacturing, and digital-native enterprises, all while ensuring data sovereignty and regulatory compliance by designing, building, owning, and operating entirely within India.

Sarvam AI’s recent $300 million funding round, featuring a $74 million investment from Nvidia, marks a watershed moment for India’s sovereign AI ambitions by scaling its GPU infrastructure from 2,000 to 10,000 Nvidia Blackwell GPUs. Partnering with HCLTech to develop a sovereign data center in Odisha, Sarvam AI aligns closely with the government-led IndiaAI Mission to bolster domestic AI capabilities and reduce reliance on foreign infrastructure, with co-founders Vivek Raghavan and Pratyush Kumar emphasizing the milestone as proof of India’s maturing capital structure for sovereign AI development.

In a landmark ₹10,000 crore to ₹15,000 crore project, Larsen & Toubro’s AI infrastructure arm LTN Compute and Vyoma.AI are deploying India’s largest single-cluster AI facility: a 10,000-GPU NVIDIA B300 mega-cluster in Chennai designed for 250 MW capacity and 150 MVA power readiness. This deployment, hailed by L&T Chairman S N Subrahmanyan as a key milestone in the 'Gigawatt AI Infrastructure Mission,' significantly elevates India’s sovereign AI capacity, enabling large-scale AI inference, fine-tuning, and training that positions the country as a global hub for next-generation AI compute.

Since rebranding to Larsen & Toubro-Vyoma in late 2025, the company has strategically shifted from merely providing data center real estate to owning and operating gigawatt-scale AI factory infrastructure in partnership with Nvidia. Their Chennai data center, designated by Nvidia for AI cloud infrastructure, hosts Blackwell GPU clusters built by E2E Networks using Nvidia HGX B200 systems and Enterprise software, running AI models across diverse sectors. By offering GPU-as-a-Service and colocation on cloud-native platforms, and leveraging advanced security technologies like Nvidia Confidential Computing through partnerships with Fortanix, Vyoma.AI is positioning itself as a sovereign, secure, and integrated hyperscale AI cloud business focused on recurring compute revenue rather than one-off projects.

Bharti Airtel's Nxtra is spearheading India's AI data center expansion by targeting a massive 1 GW of AI-ready capacity by 2030, driven by surging enterprise demand for GPU-intensive infrastructure. This growth is supported by Nxtra's strategy to build hyperscale data centers across major cities, integrating them with Airtel's extensive fibre network, subsea connectivity, and edge computing capabilities, thereby creating a comprehensive digital infrastructure platform. Sustainability remains central to Nxtra's vision, with commitments to achieve net-zero emissions by 2031 and to construct facilities adhering to LEED Gold standards, leveraging AI-driven energy optimization to minimize environmental impact.

According to OmniScience's projections, India faces an unprecedented capital challenge, requiring an estimated USD 350–435 billion by 2030 to scale AI and traditional data center infrastructure. Notably, AI-optimized data centers demand six to seven times more capital per megawatt than traditional facilities, with operational capacity needing to surge from 1.6 GW today to between 19 and 23 GW, including 7–9 GW dedicated to AI workloads alone. This stark disparity underscores the unique infrastructure and energy demands posed by AI data centers, emphasizing the need for meticulous planning to balance scale, cost, and sustainability.

Highlighting the capital intensity and scale of India's AI ambitions, L&T, in partnership with Together AI, secured a landmark ₹10,000-15,000 crore deal to deploy a 10,000-GPU NVIDIA B300 mega-cluster in Chennai by 2026. This facility, designed for 250 MW capacity and 150 MVA power readiness, exemplifies the massive energy and infrastructure requirements unique to AI-optimized data centers. L&T Chairman S N Subrahmanyan hailed the project as a milestone in the company’s 'Gigawatt AI Infrastructure Mission,' positioning India as a global hub for next-generation AI compute capable of large-scale inference, fine-tuning, and training.

By mid-2026, India's AI data center landscape has become a battleground where global hyperscalers and domestic players vie for dominance, underscoring the country's rapid integration into the global AI infrastructure race. Meta's strategic lease of a 168MW AI-ready facility from Reliance Industries in Jamnagar and Google's ambitious $15 billion investment culminating in an AI hub in Visakhapatnam exemplify how international giants are anchoring their presence in India. Simultaneously, domestic and international infrastructure providers like AirTrunk are committing staggering capital—3 trillion rupees by 2030—to scale India's digital backbone, propelling the nation's data center capacity to a projected five-fold increase reaching 12 GW, with AI-specific capacity soaring to 6,546 MW. This confluence of investments not only elevates India as a formidable AI infrastructure hub but also signals a strategic alignment where local capabilities and global ambitions converge.