India produces one-fifth of the world’s data. Yet until recently, the country operated just 3% of global data centre capacity.
Why does this gap exist? It wasn’t a technological limitation, it was economic uncertainty. For years, Indian businesses sent their computing workloads abroad, paying premium prices to access infrastructure in Singapore, Ireland, or the United States. Foreign cloud giants hesitated to commit billions despite India’s cost advantages, worried that operating data centres here could create a “permanent establishment” triggering tax liability on their global income.
On February 1, 2026, Finance Minister Nirmala Sitharaman eliminated that uncertainty. Her Budget announcement provided the policy clarity that had restrained investment for years. The response was immediate: Minister Ashwini Vaishnaw reported that total investment commitments in India’s data centre sector had accelerated from $70 billion to $90 billion, a $20 billion increase coinciding precisely with the Budget framework.
What triggered this wave? A tax holiday until 2047 for foreign cloud companies using Indian infrastructure to serve global customers, combined with transfer pricing clarity through a 15% safe harbour. The economics are now compelling: India’s construction costs are approximately 40% lower than Singapore and substantially more competitive than Ireland for equivalent data centre operations.
This week in The Mint Edge, we break down:
1. Why data centres have become critical national infrastructure, not just a tech asset
2. How AI workloads are redefining data centre design, power, and cooling economics
3. The December 2025 capital cascade that pushed commitments from $70B to $90B
4. How Budget 2026 rewrote tax and transfer-pricing rules for global cloud players
5. Where India’s data centre map is shifting, from Mumbai to the five-city framework
- Understanding Infrastructure: What Actually Is a Data Centre?
Before examining the investment boom, it helps to understand what’s being built. A data centre is a facility filled with thousands of servers, powerful computers that store data and run applications. When you stream Netflix, send a WhatsApp message, or use ChatGPT, your request travels to one of these facilities.
Think of it as critical infrastructure. Just like highways move vehicles and power grids deliver electricity, data centres process the digital economy. Every online transaction, every cloud-based application, every AI model, all depend on data centres.
Scale matters enormously. These facilities are measured in megawatts of power capacity, one megawatt can power roughly 1,000-1,200 Indian homes. The projects under construction in India operate at 100+ MW, with Reliance’s Jamnagar facility targeting 3,000 MW. That’s three gigawatts, exceeding the entire data centre capacity of the Netherlands or South Korea.
AI workloads fundamentally changed data centre economics. Training large language models like ChatGPT requires running thousands of specialized Graphics Processing Units simultaneously for weeks. A single Nvidia H100 GPU draws 700 watts under load. Eight GPUs in one server require 10+ kilowatts per rack, versus 200-300 watts for traditional cloud servers.
What does this mean practically? Legacy facilities designed for regular computing cannot simply add AI workloads. The electrical infrastructure, transformers, distribution panels, circuit protection, needs complete redesign. Cooling systems must handle 50-150 kW racks instead of 5-15 kW racks, requiring industrial-grade cooling that didn’t exist in older facilities.
This creates a structural advantage for India. All major upcoming facilities, Google’s Visakhapatnam hub, Reliance’s Jamnagar centre, Microsoft’s Hyderabad facility, are designed AI-native from the ground up. Singapore and Amsterdam face $500 million+ retrofit costs to upgrade legacy infrastructure, making new construction in India more economically attractive.
2. December 2025: The $50+ Billion Cascade
The investment wave began weeks before Budget 2026, but the timing and scale revealed careful coordination. On December 9, 2025, Microsoft CEO Satya Nadella met Prime Minister Modi in New Delhi. The same day, Microsoft announced $17.5 billion in investments from 2026 to 2029, adding to a prior $3 billion commitment. The centerpiece: a hyperscale data centre in Hyderabad set to go live mid-2026, described as roughly equivalent to two Eden Gardens stadiums combined.
One day later, on December 10, Amazon announced over $35 billion by 2030 at its annual Smbhav Summit in New Delhi. This builds on $40 billion already invested since 2010, bringing Amazon’s total India commitment to approximately $75 billion. Industry analysts called it “the fastest $50+ billion capital commitment in Asian technology history.”
These weren’t isolated moves. Google had announced its $15 billion Visakhapatnam AI hub on October 14, 2025, in partnership with AdaniConneX and Airtel. The project combines gigawatt-scale operations, a new international subsea cable gateway connecting India’s east coast to global networks, and clean energy infrastructure, all scheduled for deployment between 2026 and 2030.
The timing created market momentum. These December announcements came just weeks before Budget 2026, with growing clarity that Finance Minister Sitharaman would unveil a comprehensive tax framework. Minister Vaishnaw’s statement that commitments had accelerated from $70 billion to $90 billion, a $20 billion increase, coincided precisely with this Budget preparation period.
3. Budget 2026: What Did the Government Actually Announce?
On February 1, 2026, Finance Minister Sitharaman delivered her ninth consecutive Budget speech. The data centre announcement was precise but required careful reading to understand its full implications.
The core offer: If you’re a global cloud company, Amazon AWS, Microsoft Azure, Google Cloud, and you want significant tax benefits, you must run your global cloud services from Indian data centres through an Indian entity. In return, the government offers a tax holiday until 2047 on those cloud services.
What’s the actual incentive? If a foreign cloud company uses data centres located in India, routes its global cloud services through India, and serves Indian customers via an Indian reseller entity, then profits earned from those global cloud services will be tax-free until 2047. That’s over 20 years of tax certainty.
But there’s a critical condition many headlines missed: Infrastructure must actually be in India. Cloud workloads must genuinely run from Indian data centres, not just be billed from India. Services to Indian customers must go through an Indian reseller entity, not directly offshore. The government implemented a 15% transfer pricing safe harbour, if the Indian data centre is a related company, authorities will accept only 15% margin on costs to prevent profit shifting.
Why is India doing this? Three strategic reasons.
First, monetize India’s data centre boom. India has built massive ground capacity, but global workloads still sit abroad. This forces actual usage, not just construction.
Second, make India a global cloud hub, like Ireland or Singapore, but at much larger scale and lower cost.
Third, treat data centres as infrastructure.
The government is officially declaring: data centres equal roads, power, and ports. That’s a fundamental policy signal.
4. The Economics: Why India Wins on Cost
India’s operational data centre capacity reached approximately 1,150 MW in late 2024, according to Cushman & Wakefield research. Singapore, Asia’s established hub, operates roughly 1,300 MW. Yet India’s market trajectory is dramatic, from $9.79 billion in 2025 to a forecast $21 billion by 2031, representing approximately 14% compound annual growth.
Capital
expenditure reveals the advantage. Data shows India’s median construction cost at $6.8 million per MW versus Singapore’s $11.23 million, approximately 40% lower. Land acquisition costs remain substantially lower, particularly in tier 2 cities versus Mumbai or Bengaluru.
Power costs create ongoing operational leverage. Industrial tariffs in Gujarat or Andhra Pradesh run rupees 4-6 per unit versus rupees 12-15 in Singapore. For a 100 MW data centre operating continuously, this translates to tens of millions in annual savings. India’s coastal geography enables seawater cooling systems, reducing cooling costs relative to landlocked alternatives. Mumbai’s electricity tariff at 6.71 US cents/kWh is over 50% lower than Shanghai, further improving operational economics.
Engineering talent provides additional arbitrage at 30-40% of Singapore rates. India now has 16+ submarine cable landing points with tier-1 peering in Mumbai and Chennai, matching Singapore’s connectivity while undercutting costs.
The primary constraint remains power infrastructure. India requires 10 GW of dedicated data centre capacity by 2030, necessitating grid investments of 0.4-0.9% of GDP over the next decade. Budget 2026 addressed this partially by waiving basic customs duty on nuclear reactor imports and providing concessional financing for renewable energy integration.
Water sustainability presents environmental limits. Existing centres used 150 billion litres in 2025, projected to double to 358 billion by 2030. In water-stressed regions, this creates conflicts with agriculture and residential use. Coastal facilities can use seawater cooling, but inland facilities face genuine sustainability questions.
5. What This Menas: Why Analysts Are Both Excited and Cautious
For data centre operators, this framework is transformative: higher occupancy from global clients, long-term revenue contracts, dollar-linked revenues, and predictable tax regime until 2047. That’s why companies openly cheered the Budget announcement.
The $90 billion investment pipeline is already flowing. In just the past six months: Google, Microsoft, Tata, Reliance, Adani, and Greenko have made combined announcements exceeding this figure. The Budget policy essentially locks these investments into actual utilization rather than just construction.
However, some analysts flagged important limitations. Bernstein noted in their post-Budget analysis: “While this sounds great to hear, the eventual impact may be low given multiple complexities.”
Two critical constraints emerged. First, latency-sensitive workloads won’t necessarily move to India. Realtime applications, video calls, gaming, live AI inference, need servers close to end-users, not half a world away. So India may capture AI training, batch processing, and storage, but not all real-time workloads that require minimal latency.
Second, the U.S. CLOUD Act risk. American law allows U.S. authorities to access data from U.S. companies even if servers are overseas. This could prevent sensitive sectors, banking, defense, government systems, from using India-hosted global clouds regardless of physical location. European companies face similar restrictions under GDPR.
6. Geographic Distribution: From Mumbai Dominance to the 5 City Framework
Mumbai commands 44-54% of India’s data centre market, driven by submarine cable infrastructure. Sixteen-plus cables terminate in the metro region, providing direct global connectivity. The financial sector represents 35% of enterprise cloud demand and concentrates in Mumbai, with Navi Mumbai alone projected to reach 800 MW capacity.
Chennai holds 8-14.5% current share but is expanding rapidly, with India’s second-highest undersea cable connectivity. Proximity to automotive and electronics manufacturing clusters drives Internet of Things demand as factories deploy sensors and automation requiring low-latency computing. Real estate costs remain lower than Mumbai, attracting new entrants.
Bengaluru’s advantage lies in talent density. Over 1.5 million technology workers create demand for co-located cloud infrastructure to reduce latency for development workloads. Software engineers need rapid access to computing resources for coding, testing, and deployment. The city is experiencing 16.5% annual capacity growth.
The next frontier involves tier 2/3 cities. Visakhapatnam’s 1 GW Google-Adani-Airtel facility establishes the city as a connectivity hub serving not just India but Southeast Asia, Australia, and beyond. Multiple international subsea cables will land on India’s eastern coast, complementing west coast infrastructure.
Reliance’s Jamnagar facility targets 3 GW capacity, three times India’s current total. The site leverages Reliance’s existing massive oil refining and petrochemical complex with substantial power infrastructure. A 5,000-acre green energy complex nearby, combining solar, wind, and green hydrogen, will power the data centre entirely with renewable energy, using Nvidia’s latest Blackwell AI semiconductors.
Pune addresses 600+ million OTT streaming users requiring under 10ms latency. Edge computing capacity is projected to double from 80-100 MW to 160-180 MW by 2028, serving applications demanding real-time processing.
The Bottom Line: Infrastructure First, Not Hype Frist
This Budget made three important moves quietly.
It made India impossible to ignore as a cloud infrastructure base.
It turned data centres into a recognized national economic asset.
It chose infrastructure-first AI over hype-first AI.
The India AI Mission received measured funding: rupees 800 crore spent in FY26, rupees 1,000 crore allocated for FY27, against a total mission size of rupees 10,372 crore. The Economic Survey explicitly stated India will take a slow, measured AI path focused on use cases and infrastructure, letting private capital lead large deployments.
This won’t move all cloud workloads to India overnight. Latency requirements, data sovereignty concerns, and regulatory complexities remain genuine constraints. But it anchors long-term global cloud capital firmly on Indian soil.
India is addressing its data infrastructure gap through targeted policy, cost arbitrage, and strategic timing. From 3% of global capacity in 2024, the country is expanding to 1.7 GW by end-2026. The market is forecast to grow from $9.79 billion to $21 billion by 2031.
For investors, this represents exposure to cloud migration, AI adoption, digital infrastructure as a foundational sector, and India’s technology growth as the country moves from consuming to producing digital services.
The risks remain substantial, power infrastructure, water constraints, regulatory implementation, tier 2/3 execution challenges. What makes this moment significant is not the absence of challenges but the alignment of policy, economics, and timing.
Whether the $90 billion promise delivers will depend on execution over the next 2-3 years as projects move from announcement to operation. The infrastructure being built today will determine India’s position in the global digital economy for the next decade.
Disclaimer
This article is for informational purposes only and does not constitute investment advice, a recommendation, or an offer to buy or sell any securities. The views expressed are based on publicly available information and industry sources as of the date of publication. Market conditions, regulations, and assumptions may change over time. Readers should conduct their own independent analysis and consult qualified financial advisors before making investment decisions. The author and publisher accept no liability for actions taken based on this content.
