₹30,000 Crore For 400 MW; Inside Adani’s Massive Google AI Data Centre Bet

Adani Enterprises Ltd (AEL) has disclosed financial details for the first time for its proposed 1-gigawatt artificial intelligence (AI) data centre hub. Adani Group Chief Financial Officer Jugeshinder Singh outlined funding and cost estimates for the project during the July 29, 2026 first-quarter earnings call.

"We are roughly around, say, about Rs 70 crore to Rs 75 crore per megawatt in terms of construction," Singh said.

The project targets an annual dollar-equivalent return on investment of approximately 12 per cent. Applying the upper-end estimate of Rs 75 crore per megawatt to the initial 400-megawatt phase implies a construction cost of around Rs 30,000 crore. These figures set a baseline for investors.

Scope and Timeline

AdaniConneX, a joint venture between Adani Group and global data centre operator EdgeConneX, is executing the 400-megawatt order. This serves as the first phase of Google's 1-gigawatt AI hub. 

Singh confirmed that the 400 MW represented the portion currently under execution and said "other parts" of the project would be implemented subsequently.

An AdaniConneX spokesperson, "The recently announced 400 MW order represents the first awarded phase for AdaniConneX under the broader 1 GW Google AI Hub in Vishakhapatnam, one of India's most significant AI Infrastructure developments.

This will help to unlock India's massive potential in the AI age. AdaniConneX is proud to partner with Google on this historic project, which will help define the future of India’s digital landscape.

Visakhapatnam is set to become a global technology destination, and we are thrilled to be the architects of this monumental journey."

The project spans three campuses. Adani and Google announced the 1-gigawatt hub in October 2025, which they are co-developing with Bharti Airtel's Nxtra.

The October 2025 announcement stated that Google and Adani would co-invest in transmission infrastructure, renewable energy generation and battery energy storage systems supporting the project.

Construction formally commenced at the Tarluvada site in Andhra Pradesh in April 2026.

Full construction and customer deployment will require 40 to 48 months. The initial 400-megawatt capacity will become operational within 2.5 years. The end customer will supply the graphics processing units. This agreement excludes one of the most expensive infrastructure components from the current construction cost estimate.

Financing and Capital

Project finance drives the strategy. The company structured the funding as project finance rather than conventional property funding.

It targets debt tenors of around seven years, backed by service contracts offering roughly 10 years of revenue visibility.

"We are looking at a more traditional infra-type funding, and we are looking for this business to track close to investment grade," Singh said.

A 70 to 80 per cent debt ratio on an estimated Rs 28,000-crore to Rs 30,000-crore construction cost implies an equity or advance funding requirement ranging between Rs 5,600 crore and Rs 9,000 crore.

AEL completed a Rs 15,000-crore qualified institutional placement on July 7, 2026. However, the offer document does not explicitly earmark proceeds for the Visakhapatnam project.

Singh said that hyperscale customers do not automatically yield higher returns, as yields depend on individual contracts. He maintained that the "overall economics are very good". 

During AEL’s Q3 FY26 earnings call in February, management had declined to discuss project economics, and the subsequent Q4 FY26 earnings call in April yielded little additional information.