Data Centers Boom in India: Why Infrastructure Due Diligence is Becoming a Deal Breaker

May 05, 2026

India's data center sector has crossed the threshold from emerging opportunity to institutional asset class. Data center investment in India is no longer exploratory, hyperscale commitments are now anchor bets on a market that is structurally transforming.

Yet beneath the headline numbers, a more complex reality is taking shape: the gap between announced megawatts and operational megawatts is widening, and the infrastructure constraints responsible for that gap are determining which deals close, which stall, and which quietly destroy capital. For infrastructure funds evaluating India, technical due diligence in infrastructure has stopped being a checkbox exercise. It is now the difference between a performing asset and a stranded one.

The Scale of Capital and the Urgency Behind it

India attracted approximately $14.7 billion in data center investments between 2020 and April 2025, with foreign institutional investors contributing nearly 86% of the total. As per an industry report, cumulative investment commitments reached $126 billion by end-2025, expected to surge a further 45% to exceed $180 billion in 2026.

Additionally, Microsoft pledged $17.5 billion, AWS committed $35 billion, and Google $15 billion, collectively, $67.5 billion from three hyperscalers. Domestically, TCS partnered with TPG to build the $2 billion HyperVault AI data center platform. Yotta Data Services is deploying one of Asia’s largest NVIDIA Blackwell HGX B300 superclusters with over 20,000 NVIDIA Blackwell Ultra GPU.

The India data center market, valued at $5.55 billion in 2025, is forecast to reach $13.11 billion by 2034 at a CAGR of 10.01%. Between March 2025 and April 2026, operators announced roughly 30 large projects collectively adding 3.5 GW of planned capacity. On paper, the story is compelling. But in practice, it is increasingly an execution story, and execution risk is where investor returns are made or lost.

Power: The Gating Variable No Announcement Can Bypass

If there is one factor that will determine how quickly India's data center ambition converts into operational capacity, it is power, not land, not demand, and not financing. Grid availability is the constraint that neither government commitments nor developer representations can substitute for. Data centers are expected to consume approximately 3% of India's electricity by 2030, up from less than 1% currently.

The shift from conventional cloud workloads to AI inferencing has fundamentally altered power density requirements: where a traditional hyperscale facility operated at 5–10 kW per rack, AI-optimised facilities are designed around 30 kW/rack or higher, with next-generation deployments targeting well beyond that.

The financial consequences of getting this wrong are quantifiable. One month of construction delay on a 60 MW facility costs an estimated $14.2 million in foregone revenue and $1.8 million in monthly interest carry. A three-month delay compresses development IRR from approximately 17.1% to 12.6%, eliminating the return premium that justified development risk over investment-grade alternatives. There is a documented case of a developer who received news, three months after closing, that state electricity board infrastructure would not be available for 18–24 months. The "opportunity" had become a INR 15 crore liability. This scenario is not uncommon in India's faster-growing emerging markets.

When it comes to infrastructure funds, power technical due diligence in infrastructure cannot be based on developer assertions alone. Verification should ascertain that there is available capacity within the nearest 220/400 kV substation, last mile transmission facilities, bankable PPAs with IPPs, and alternative grid connections. Verbal commitments from state electricity boards are not bankable, and in India's fragmented grid landscape, the gap between commitment and delivery has real capital consequences.

Permitting: Thirty Clearances, Fragmented Across States

India's data center policy environment is, in practice, a "one nation, many laws" reality. The central government's direction, infrastructure status designation, the DPDP Act 2023, and the Draft National Data Centre Policy, provides a strategic framework. Implementation is state-specific, with divergence in incentive structures, clearance timelines, and regulatory maturity creating deal risk that is easily underestimated from a distance.

Facilities exceeding 20 MW, which now represent more than half of India's installed capacity, typically require up to 30 separate clearances spanning land use, utilities, environment, and security. This process can extend timelines by several months for large campuses.

States like Tamil Nadu and Telangana have moved toward time-bound single-window clearances; Andhra Pradesh's Data Centre Policy 4.0 (2024–29) explicitly targets 200 MW of additional capacity with dedicated fiscal support. But these are exceptions, not the baseline. For every data center infrastructure investment, the actual clearance workflow for that specific site, not the state's published policy, must be mapped, and worst-case permitting timelines explicitly built into the financial model.

The AI Workload Shift: A Different Infrastructure Thesis

The character of demand has shifted in ways that make pre-AI-era infrastructure assessments unreliable. Enterprise colocation once defined the market at sub-10 MW requirements. Hyperscalers raised that to 25-50 MW. With AI workloads, per-deployment requirements are climbing toward 75-100 MW, with fundamentally different infrastructure specifications. AI data centers demand power redundancy architectures, thermal systems capable of managing high-density GPU clusters, direct-to-chip and immersion liquid cooling, and network latency profiles that general-purpose builds were not designed to deliver.

Facilities in established markets like Mumbai and Noida that were designed as general-purpose hyperscale builds are deploying AI workloads as dedicated zones rather than core design principles. That gap matters for long-term tenancy stability. Hyperscalers evaluating colocation decisions are increasingly screening for AI-readiness at the infrastructure layer, and facilities that cannot demonstrate it face tenant attrition risk. Technical due diligence in infrastructure should assess rack power density design targets, cooling architecture, PUE ratios, water availability, and the facility's capacity to accommodate high-density GPU lifecycle management without structural retrofits.

The Diesel Dependency and the ESG Liability

India's data center sector carries what industry observers have candidly described as a structural liability: deep reliance on diesel generator farms for backup power. A 100 MW hyperscale campus requires backup power infrastructure of its own, with substantial diesel storage. This is an environmental, regulatory, and commercial risk converging simultaneously. Policymakers are tightening emissions norms. Global hyperscalers operating under 100% renewable energy commitments are auditing their colocation partners. The Bureau of Energy Efficiency has introduced national PUE optimization guidelines. India achieved nearly 50% non-fossil fuel generation capacity by mid-2025.

The financial stakes are direct. ESG-compliant facilities command 15–25% premium lease rates. Facilities with high diesel dependency face not only regulatory risk but also softening tenant demand from the largest and most credit-worthy hyperscalers. Any infrastructure fund conducting technical due diligence must independently assess the renewable energy transition pathway, open access power arrangements, and the bankability of long-term green power offtake agreements.

What Infrastructure Funds Must Do Differently

India's data center infrastructure investment opportunity is structurally sound. The demand drivers, data localization under the DPDP Act, AI compute requirements, sovereign cloud frameworks, and a billion-subscriber internet base, are durable and policy-backed. Announced projects worth $60–70 billion are in the pipeline over the next five years. The market's trajectory is not in question. What is in question is which assets in that pipeline will deliver returns commensurate with risk, and which will compress IRR through execution failures that a rigorous diligence process would have identified in advance.

The investment approach must evolve accordingly. Technical due diligence in infrastructure, covering grid access and redundancy, permitting workflow and status, AI-readiness of power density and cooling architecture, renewable energy transition credibility, and geographic-specific infrastructure maturity, is not a secondary workstream to be completed after commercial terms are agreed. It is the primary risk-adjusted return analysis. Infrastructure milestones should be conditions precedent to closing, not post-close undertakings. A minimum 20% infrastructure contingency should be applied to any project requiring new substation construction or fiber build-out.

The deals that will define India's digital infrastructure decade will not be won on announced megawatts. They will be won by investors who can independently distinguish between capacity that is planned and capacity that is actually deliverable, on time, at the right power density, and with the infrastructure depth that AI-era tenants require.

In India's data center race, capital is abundant. The constraint is execution and the investors who master infrastructure due diligence will define who captures the returns.