Companies have pledged more than $126 billion to build AI data centres in India. Almost none of that money is going to the government’s own AI scheme. It’s private infrastructure, built by corporations racing to host India’s AI boom.
mcqquestion.com
Science & Technology · SciTech0041
AI Infrastructure in India
$126B+ committed by private players, apart from the government’s IndiaAI Mission
$17.5B
Microsoft — largest Asia investment
$100B
Adani — 5 GW target by 2035
$15B
Google — Visakhapatnam AI hub
$12.7B
AWS — through 2030
$6-7B
TCS HyperVault — 1 GW target
168 MW
Meta-Reliance, Jamnagar (initial)
APAC’s 2nd-largest DC market
1.6 GW operational
+26.3 GW demand by 2031-32
Two tracks, one ecosystem
The government’s IndiaAI Mission subsidises compute access for smaller players. Private hyperscalers build the physical data centres much of that compute runs on — two complementary tracks, not competing ones.
Must Know
A Private Race for Compute
What Is a Data Centre
- BasicsTraining and running AI models needs huge banks of specialised chips, called GPUs, housed in purpose-built facilities called data centres. A “hyperscale” data centre is one built at massive scale, usually to serve cloud or AI workloads for many customers at once.
A Private Buildout, Separate From Government
- ScaleGlobal technology companies are now investing heavily in Indian hyperscale capacity, separate from any government scheme. Cumulative private investment commitments for AI data centres in India now exceed $126 billion.
- DistinctionThis private buildout is distinct from the IndiaAI Mission, the government’s own scheme to subsidise GPU access for Indian startups and researchers. See SciTech0033 — The IndiaAI Mission for that scheme’s own structure, and SciTech0044 — AI Centres of Excellence in India for its research and skilling side.
Good to Know
Who’s Building What
India’s Data Centre Scale
- ScaleIndia is the Asia-Pacific region’s second-largest data centre market, with 1.6 gigawatts (GW) of operational capacity. A further 3.1 GW is under construction or planned, and over 10.5 GW more is at the land-acquisition stage.
Named Investors, and TCS HyperVault
- DetailMicrosoft has committed $17.5 billion to Indian cloud and AI infrastructure, its largest investment anywhere in Asia. Amazon Web Services has committed $12.7 billion through 2030, with $8.3 billion of that in Maharashtra alone.
- DetailThe Adani Group has pledged $100 billion toward a national AI infrastructure platform, targeting 5 GW of capacity by 2035. Google is investing $15 billion by 2030 in a gigawatt-scale AI hub at Visakhapatnam, Andhra Pradesh, its largest AI hub outside the United States.
- DetailTata Consultancy Services launched “HyperVault,” targeting 1 GW of AI data centre capacity, with $6-7 billion invested over five to seven years. A TCS-TPG joint venture is putting in $2.03 billion of that; OpenAI is HyperVault’s first customer.
Current Affairs
📰 Current Affairs
The Mission’s Underspend
- 6 Aug 2026Parliament’s 31st Standing Committee Report flagged that only 32% of the IndiaAI Mission’s 2025-26 funds were used. Fellowship targets also fell short: just 150 undergraduate fellows were selected, against a target of 5,000. The Mission’s 2026-27 budget was halved by the Finance Ministry.
- SourceThe Print.
Environmental Clearance Rules
- 5 Aug 2026The government clarified that most AI data centres don’t need standalone Environmental Clearance. Larger projects still need it: any built-up area over 20,000 square metres, or township-style projects over 50 hectares.
- DetailIndia’s data centre capacity has grown more than fourfold in six years, from 375 MW in 2020 to about 1,575 MW today.
- SourceThe Print, Tribune India.
The Mission’s Structure, and The Jamnagar Deal, Summarised
- 27 Jul 2026The IndiaAI Mission, approved March 2024 with an outlay of ₹10,371 crore across 7 pillars, offers subsidised GPU access at about ₹65 per GPU-hour through 15 empanelled providers. Its Safe & Trusted AI pillar has approved 13 Responsible AI projects, including Saakshya, a deepfake-detection framework built by IIT Jodhpur and IIT Madras.
- SourcePIB.
- Jun 2026Meta and Reliance Industries announced India’s first AI-enabled hyperscale data centre, at Reliance’s campus in Jamnagar, Gujarat — 168 MW initial capacity, cooled with desalinated seawater instead of fresh water.
- Full coverageCA0018 — Reliance and Meta Partner to Build India’s First AI Hyperscale Data Centre.
✅ Test Yourself
Question 1 of 1
Great to Know
The Real Resource Costs
Two Complementary Tracks
- ReframeGovernment and private infrastructure play complementary roles. The IndiaAI Mission subsidises access to compute for smaller Indian players. Private hyperscalers build the physical data centres that much of that compute ultimately runs on.
The Real Cost: Electricity
- StakesThis scale of buildout carries real resource costs. AI data centres in India are projected to add 26.3 GW of new electricity demand by 2031-32. The sector’s own power load has already grown fast, from around 1,400 megawatts in 2024 to an estimated 1,825 megawatts in 2026.
The Real Cost: Water
- TensionWater is a growing concern too. More than half of India’s data centres sit in water-stressed regions. Visakhapatnam, the site of Google’s new hub, is itself in a district with some of Andhra Pradesh’s lowest groundwater availability.
- ResponseFacilities are responding with technology choices, not just siting. Meta’s Jamnagar data centre, for instance, uses desalinated seawater for cooling instead of fresh water.
Was this helpful?
Leave a Reply