India stands at third place on Stanford’s Global AI Vibrancy Index after the US and China. Amazon ($35B), Microsoft ($17.5B), Google ($15B) dropped into Indian cloud infrastructure. MoUs are signed across the states to have the data centers, AI hubs or Centers or GCCs. We could possibly be the world’s AI back office in the near future.
In the middle of all this, lets focus only on our priority. According to the most recent Agriculture Census and NITI Aayog reports (2025), 86.2% of Indian farmers are classified as small and marginal farmers. Plots are smaller than 2 hectares. With respect to the digital adoption in agriculture remains or stays stubborn 20%.
So. Is any of this AI or tech or robotics actually reaching the field? Or is it a conversation or LinkedIn talks happening on it? Some clarity or contradictions or even paradoxes. That’s what I’ve been trying to figure out by putting the thoughts around it.
#The water paradox
The smart farming helps the farmers to reduce the inputs usage, reduce the water and more. AI powered precision farming makes it more accurate and reliable.
The data centers running those AI models need 10x more water (TBP, fresh water) than conventional ones. The videos has been surfing on issues raised on drinking water in other countries (I wish them to be AI generated) .
Obviously, everything comes at a cost. No technology is cost-free. The costs just move around different resources available.
#Climate change
Indian agriculture is facing disruptions at once. Climate change has been all around. This has been breaking the predictability farmers have relied on for generations. And yes, AI is promising to replace that lost intuition with data. Some reports are pretty good on weather and it appears to be the promising tool. Not hype. Maybe a genuine gap being filled.
#AI in Indian Agri
Bharat-VISTAAR: Government of India launched January 2026. Multilingual AI advisory integrating AgriStack (check out the portal) records with ICAR validated practices. Speaks Tamil, Telugu, Hindi and more. Maybe more enhanced with Sarvam ai? Just a thought.
Some companies like FASAL, Cropin, BharatAgri, Kissan AI, DeHaat, AgroStar FarmerChat, and more work their best to deliver the best solutions to the farmers in the field of Precision Farming, Data Intelligence, IoT devices, devices. Democratizing expert knowledge through a AI bot in the regional would be more successful for the farmers.
Soil moisture monitors, crop health trackers, weather stations at farm level, AI bot, robotics (Covered some of this in #DecodeAgri22. State of play for Indian context specifically). Feeding these real-time data into advisory platforms makes the recommendations are getting good. Who owns that data is a question we’ll get to.
#The data problem
The unglamorous foundation of all of this state is that data doesn’t exist yet in usable form.
India’s agricultural data is scattered. In photos, PDFs, paper records. Most of them has to manually scraping, structuring, and translating this before they could train anything useful on it.
“It’s all over the place, a lot of it is not documented well… you actually have to make an effort.” — Rhishi Pethe, Metal Dog Labs
#Whose data is it anyway
The Government of India has agricultural data portals. Hoping that AgriStack might be of some help. But having a portal isn’t the same as farmers understanding what’s being collected, who can access it, or who profits from it.
India built UPI. India built Aadhaar. The question is whether we build the data rights framework established before the extractions economy goes out of hands. Lets see.
#Whose food system is this actually
Here’s the question I think is being not asked much.
Look at where the energy around startup, youth and the investor’s attention in the field of Indian agri-tech right now even from the other countries. Organic farming, A2 cow milk, Millets for urban consumers, Exotic fruits, Hydroponic (strawberries, cabbage, greens and lettuce), Aeroponic herbs are the eye of the current system. These controlled environment produces hit the export quality (both MRLs & MLs), lands in European supermarkets.
These systems actually work. Precision farming reduce pesticide use dramatically and produce consistent quality. It is moving towards high end value, better margins and revenue. What about the cereals, oilseeds, cash crops?
#We’ve been here before actually
The Green Revolution help the entire nation towards maximizing the yields of wheat and rice. It fed hundreds of millions. It also left everything else behind. Pulses, millets, oilseeds, diverse cropping systems. Decades later, the rice-heavy diet is being linked to metabolic issues including diabetes.
Now the pendulum is swinging. Protein is the new obsession right now. Exotic horticulture is where the investment goes. Fibre hasn’t had its moment yet but will rise.
Carbs, Proteins, Fats, Vitamins and Fibre! The body doesn’t pick one. Why should the food system?
We can’t now hype protein and ignore the carbohydrate base that still feeds the majority of this country. The PDS distributes cereals and millions of people are covered under food security schemes.
#Millets & missing middle
Millets are having a cultural moment nutritionally, ecologically, and for smallholder farmers who grow them.
But the millets come from open fields, small plots, irregular edges, variable soil and adverse climatic conditions with minimum requirement of water. The same AI tools transforming controlled-environment horticulture don’t automatically transfer here.
There’s a risk in the current millet moment. Trying to converting them into exotic. Selling ragi flour at ₹400 a kilo in a wellness store is not the same as normalizing millets across India’s food system. The goal should be making them affordable and accessible. Not as premium product for people who can afford to think about ancient grains.
Improving millet farming, processing, and value chains so the crop is accessible at the bottom of the pyramid. That’s the actual and tough work.
Agroecology can hold this diversity and can preform better. Millets alongside legumes, alongside trees, alongside open field cereals. But agroecology doesn’t scale the same way a precision farming unit does. Diversity is the whole point. You can’t expect the same clean ROI on a polyculture system as on a controlled environment strawberry tower.
That’s the tension/chaos .
#The FPO angle
Every time a new agri-tech tool is announced, question always points towards “How small and marginal farmer can actually afford this?
FPOs changes the game here. 100 to 500 to even 1000 small farmers pooling resources. Per-farmer cost of shared sensors, drone, advisory subscriptions gets reduced. Collective bargaining with buyers, input suppliers, tech companies. Some companies target the FPOs (Rakshak) while some on rent basis. These model helps in collective certification processes too if it organic or natural produce.
#The middleman upgraded
The narrative that AI cuts out the middleman is appealing yet not clear. Platforms like NinjaCart streamline supply chains faster, leaner, less post-harvest loss. Yet, they serve as middleman but with system and laptop now. Aggregators don’t disappear when technology arrives. They just get upgraded. The system needs to ensure that whether the farmer gets more.
#Digital divide less scary than it looks
Smartphones have reached villages and UPI works where there are no bank branches. The access gap from five years ago has closed faster than expected. The gap on adopting the farmer’s app still uneven.
MIT research identified 18 explicit barriers to adoption, with a Trust Paradox at the center. Rural communities often see AI as a foreign black box. Until it feels local and grounded, adoption stays uneven. Spiky, as Omnivore’s Mark Kahn puts it.
#Policy decides whether it works or doesn’t
AI can help a farmer grow a better crop. Whether that farmer actually benefits depends on procurement prices, input subsidies, credit access, whether crop insurance actually pays out when the floods hits. These are policy decisions and the gap is bein noticed
“Monetizing farmers in the global south is next to impossible to do.” — Seamus Tardif, Myca
#How do we know if its working
Yield per hectare is the easy metric. Apart from that, soil health over five years, increased income, water use, crop & microbial diversity, affordability on the consumer level.
#Staying updated
The infrastructure is being built. Home-grown tools like FASAL, Cropin, BharatAgri, Kissan AI, DeHaat, AgroStar FarmerChat and more are genuinely interesting work happening with 20% digital adoption in agriculture. Staying updated with AI requires more time.
“The real victory won’t be in the size of the model, but in the ability to finally make it work for the person in the field.” — AgFunderNews, 2026
The person in the field is not always growing strawberries. Sometimes they’re growing other crops like soyabean or sesame. And that farmer deserves the technology too.
Cheers
Check out the similar posts: #DecodeAgri22: The War on Weeds! & other Agri posts
Previous post: Connecting the dots
If you like what you see and wish to support my work, then
Share your thoughts/suggestion at the comment section or mail at
randomwhyss[@]gmail[dot]com
Don’t miss out! Get notified about new blog posts straight to your inbox !
(No spam, pinky promise!)
Enter your mail to receive updates
Random Whys 