On this page
Adani targets up to ₹30,000 crore for 400 MW Google AI data centre phase
Our read
The most investable second-order effect is a multi-year surge in power, cooling, electrical-equipment and construction demand around Visakhapatnam, with local retail benefits lagging commissioning rather than appearing at financing close.
For operators
The planned 400 MW Google AI data-centre phase signals growing regional cloud and AI capacity that could improve the infrastructure available for retail automation, analytics and digital commerce.
Watch
Google confirmation of a binding 1 GW campus commitment and the contracted capacity for phase one.
The report,
Adani Enterprises detailed funding for AdaniConneX’s 400 MW first phase of Google’s 1 GW AI hub in Visakhapatnam, requiring up to ₹30,000 crore. The project targets 70-80% debt financing, with operations expected within 2.5 years.
Newer report , , Mint : Adani Enterprises valuation puts focus on airport retail monetisation
One email each morning: the day’s top moves in Indian retail, why each matters and what to watch. Free. Stop any time.
Reported figures
From the report. Source details below
| ₹15,000 crore AEL QIP completed July 7, | 2026 |
|---|
Also in the report
- ₹70-75 crore construction cost per MW
- ₹28,000-30,000 crore estimated initial-phase construction cost
- approximately 12% annual dollar-equivalent ROI target
- around 10 years of service-contract revenue visibility
- ₹5,600-9,000 crore estimated equity or advance-funding requirement
Other figures
- around 7-year debt tenor
Why it matters to operators and investors
The project creates potential partnership opportunities across power, cooling, connectivity, cloud services and AI-enabled retail technology as Visakhapatnam develops into a major data-centre hub.
What to watch next
- Announcement of debt arrangers, financing close, interest-cost terms and any Google/customer advance-funding contribution.
- Andhra Pradesh approvals for land, power, transmission, water and environmental clearances.
- Power-purchase agreements and renewable-plus-storage commitments sufficient for high utilisation of 400 MW.
- Orders or vendor awards for electrical equipment, cooling systems, generators and fibre connectivity.
- Construction start date, grid-energisation milestones and evidence that the stated 2.5-year operating target remains intact.
Likely next moves
Our read of what comes next — analysis, not reported by the source.
- AdaniConneX is likely to seek binding long-term capacity commitments, lender mandates and potential customer advance payments to support the targeted 70-80% debt structure.
- The company will need to secure transmission connectivity, firm renewable-energy supply, backup generation and water/cooling approvals before major construction commitments.
- Equipment procurement should favour transformers, switchgear, UPS systems, generators, liquid-cooling infrastructure, networking, fibre and specialised civil contractors.
- Google may deepen regional cloud, AI and enterprise partnerships, increasing demand for local systems integrators, telecom capacity and data-security services.
- Visakhapatnam developers and organised retail operators may target workforce catchments with convenience retail, food outlets, rental housing and mobility services.
The counter-case
The case against this reading — not reported by the source.
The headline may overstate certainty: this is a financing plan for a first phase of a proposed Google hub, not evidence of a fully contracted, funded 1 GW campus. A ₹28,000-30,000 crore 400 MW build with 70-80% leverage creates substantial refinancing, interest-rate and completion-risk exposure before utilisation is proven. AI data-centre demand can be volatile, while power availability, transmission build-out, water/cooling constraints, equipment lead times and permitting in Visakhapatnam could delay the stated 2.5-year timeline. If Google’s capacity commitments, pricing, or ramp schedule are softer than assumed, AdaniConneX could carry expensive underutilised infrastructure and face lower-than-expected returns.
The source
Published
First seen