AirTrunk Commits $30B to 5GW India Data Centers

Blackstone-backed AirTrunk pledges $30 billion and 5GW of AI data center capacity in India by 2030 - more than triple the country's current total installed base.

AirTrunk Commits $30B to 5GW India Data Centers

Blackstone's data center operator just committed more capital to Indian AI infrastructure than most countries have deployed in their entire cloud history. Now comes the hard part: actually building it.

AirTrunk, backed by Blackstone and Canada's CPPIB pension fund, announced June 5 a commitment to build more than 5 gigawatts of data center capacity in India by 2030. CEO Robin Khuda flew to New Delhi to formalize the pledge directly with Prime Minister Narendra Modi. India's current installed data center capacity sits at roughly 1.5GW. AirTrunk is committing to more than triple that figure, from a single operator.

Key Numbers

MetricValue
Total commitment$30 billion by 2030
Target capacity5+ GW new capacity
Flagship project3GW campus, Raigad, Maharashtra
Existing pipeline600MW across Mumbai, Chennai, Hyderabad
Market entryApril 2026 via Lumina CloudInfra acquisition
India's current base~1.5GW total data center capacity
India's 2030 projection~8GW (industry estimate)

The Scale of the Bet

What 5GW Actually Means

Five gigawatts of data center capacity is a figure that doesn't map neatly to intuition. India's entire data center sector today handles about 1.5GW. AirTrunk's commitment, combined with the announcements from Amazon ($35B), Google ($15B), Microsoft ($17.5B), and domestic players like Adani (reportedly $100B through 2035), would collectively require India to build more capacity over the next four years than it has constructed in its entire history.

Deloitte estimates that Asia-Pacific data centers alone will require tens of additional terawatt-hours of electricity by 2030. India's share of that demand is growing fast: AI-related colocation leasing doubled to 348MW annually last year, now representing nearly 20% of total demand, according to industry data.

India's Starting Point

The government has moved to pull this investment in. Foreign tech firms operating AI services from Indian data centers qualify for a 20-year tax exemption running through 2047. The IndiaAI Mission is backed by about $1.2B, the India Semiconductor Mission by roughly $9B. These programs don't build power lines or lay fiber, but they signal that New Delhi is treating AI infrastructure as a strategic priority, not a permit backlog.

Robin Khuda stated the investment thesis rests on three conditions: government support, a large pool of technical talent, and access to renewable energy. PM Modi called the commitment "among the largest made to India's digital infrastructure sector."

AirTrunk founder and CEO Robin Khuda meets Prime Minister Narendra Modi in New Delhi on June 5, 2026 AirTrunk CEO Robin Khuda met PM Modi to formalize the $30B commitment on June 5, a rare direct diplomatic signal from a private infrastructure operator. Source: socialnews.xyz (IANS/PMO)

Inside the Infrastructure Plan

The Flagship Raigad Campus

The largest single piece of the commitment is a 3GW data center campus at the Raigad Pen Growth Center in Maharashtra, just outside Mumbai. Maharashtra Chief Minister Devendra Fadnavis confirmed a signed letter of intent for land allotment in early June. Industry estimates put the Raigad investment at about $21B - around 70% of the total program.

Three gigawatts in a single campus is truly unusual at a global level. For comparison, most announced hyperscale campuses top out in the 300-500MW range for a single contiguous site. AirTrunk is planning 6-10x that at one location.

The Existing Footprint

AirTrunk entered India in April 2026 by buying Lumina CloudInfra, picking up a 600MW development pipeline that spans India's three largest data center markets. The site identifiers show what's already in development:

BOM1, BOM2, BOM3  -  Mumbai
MAA1              -  Chennai
HYD1              -  Hyderabad

These aren't yet operational - they're sites in development. But they give AirTrunk a geographic spread from day one, bypassing years of land acquisition and regulatory groundwork that greenfield entrants would need.

Racing Against Rival Commitments

AirTrunk is entering an already crowded field. The India AI infrastructure race has drawn both global hyperscalers and domestic conglomerates with advantages a foreign operator can't easily match.

CompanyCommitmentNotes
AirTrunk$30B, 5GW by 2030Infrastructure operator; Blackstone + CPPIB
Amazon$35B by 2030Cloud and AI infrastructure
Microsoft$17.5BData center regions already opened
Google$15B, 2026-20301GW campus broke ground in Vizag, April 2026
Adani Group~$100B through 2035Domestic; deep state-level relationships
Reliance IndustriesUnspecifiedDomestic operator, existing infrastructure

Google broke ground on its Visakhapatnam campus in April as part of the broader wave of pledges that began at India's AI Impact Summit. Blackstone is also running a separate AI compute infrastructure bet: a $5B joint venture with Google to sell TPU-based compute-as-a-service. AirTrunk is the physical layer of that broader strategy.

AirTrunk's SGP1 data center facility in Singapore - the kind of hyperscale infrastructure the company is planning to scale massively in India AirTrunk's Singapore hyperscale campus. India's planned facilities will eventually dwarf it. Source: airtrunk.com

Where It Falls Short

Power: The Real Bottleneck

Five gigawatts of data center capacity at a PUE of 1.3 draws roughly 6.5GW from the grid. India's power sector is improving, but large industrial loads - which is exactly what a 3GW campus in Raigad qualifies as - require dedicated transmission infrastructure that can take years to permit and construct. AirTrunk's discussions with government officials focused explicitly on reliable and cost-effective power. The fact that this was a conversation topic, not a solved problem, is worth noting.

Grid quality also varies sharply across the states where AirTrunk holds development pipeline. Hyderabad and Mumbai have fairly stable industrial power. Other expansion sites in the commitment are still being identified.

Water and Land

Data centers require cooling water at scale. Availability varies by region and season, and several of India's major tech cities - including Chennai - face periodic drought stress. AirTrunk hasn't disclosed water sourcing plans for the full 5GW program, only for the Raigad campus where Maharashtra's government has committed to supporting infrastructure access.

Land acquisition at the scale needed for gigawatt-class campuses is also a multi-year process in most Indian states, even with government support.

The Four-Year Clock

AirTrunk closed the Lumina CloudInfra acquisition in April 2026. The $30B commitment runs to 2030 - less than four years to go from 600MW of inherited pipeline to 5GW of operational capacity. The construction approaches that Meta is using with tent-based modular builds can compress construction timelines meaningfully, but nothing currently demonstrated gets 4.4GW of net new capacity across multiple states into operation in 46 months.

This doesn't mean the commitment is hollow - it means the $30B figure likely represents a multi-year program ceiling rather than a fixed delivery guarantee. The SoftBank €75B commitment to French data centers drew the same kind of scrutiny when announced. These numbers describe what an infrastructure program could absorb at scale, not what will be operational by a specific date.

Local Competition

Adani Group's $100B commitment isn't a foreign company learning Indian permit processes - it's a domestic conglomerate with grid connections already in place, land banks bought years ago, and relationships at the state level that move approvals faster than any foreign operator can replicate. Reliance has comparable structural advantages. AirTrunk's operational expertise and international capital access are genuine differentiators, but they don't override the structural advantages that local players hold in an infrastructure race where land, power, and permits are the binding constraints.


"Capital is mobile, and India is creating the conditions for it to thrive," Khuda said at the announcement. Getting capital there's the easy part. Getting 5GW of operating data centers by 2030 is a different problem - one that hasn't been solved at this scale anywhere in the world.

Sources:

Sophie Zhang
About the author AI Infrastructure & Open Source Reporter

Sophie is a journalist and former systems engineer who covers AI infrastructure, open-source models, and the developer tooling ecosystem.