The Cloud and the Cultivator — Andhra Pradesh's Agrivoltaics Model

Google is investing $15 billion in what is set to be Asia's largest AI data centre at Visakhapatnam, operational by 2028. A facility on that scale needs an enormous, uninterrupted supply of electricity — around 24 GWh a day, which is what a mid-sized city consumes.

That single number is where the policy problem begins.

• Draw that power from the existing grid, and it competes directly with farmers who depend on subsidised agricultural electricity.
• Meet it through coal, and the environmental cost undoes the point of building clean digital infrastructure.

So a third path has been proposed — agrivoltaics. Solar panels are mounted high above farmland, roughly 10 to 16 feet off the ground, leaving enough clearance for crops to grow and machinery to move beneath them. The land does two jobs at once: food below, electricity above.

Crucially, Andhra Pradesh does not need a new law to try this. Its existing Deemed Distribution Licence (DDL) policy already permits data centres to buy power directly from private producers, bypassing the state distribution company. That one policy instrument is what turns agrivoltaics here from a good idea into an implementable one — because it creates a buyer.

The old development model is familiar. Government acquires the land, pays one-time compensation, and the farmer exits the story — usually with a grievance that outlives the cheque.

This model inverts that. Farmers retain their land titles. What they part with is a usage right, not ownership. In return they become co-owners of the energy enterprise itself.

The structure works like this:

• Farmers are organised into solar cooperatives on the Amul model — collective ownership, collective bargaining.
• A Special Purpose Vehicle (SPV) holds the project, and farmers hold equity in it through their land contribution.
• Financing comes from a blended stack: Google's low-interest long-term debt, PM-KUSUM capital subsidy, and commercial loans backed by a Credit Guarantee Trust Fund.

Income then arrives through four streams rather than one:

1. Traditional crops grown beneath the panels — existing income preserved, not replaced.
2. High-value shade-loving crops such as turmeric, ginger and leafy vegetables, which the partial shade actually suits.
3. Value-added processing of that produce.
4. Recurring electricity sales to Google under a long-term offtake agreement.

There is arguably a fifth: savings on diesel and grid electricity for irrigation, since generation now sits on the farm itself.

ICRIER's pilots in Rajasthan, Odisha and Madhya Pradesh have shown farm income rising eight to ten times under this configuration.

The proposal envisages 5,000 systems of 1 MW each — 5,000 MW in total, at roughly ₹4 crore per MW, or about ₹20,000 crore of investment. Almost all of it private. The government's role is to enable, not to fund.