AI, satellites, drones and climate-resilient genomics dominated the technology agenda at Delhi. The headline figure is arithmetically sound — and rests on an assumption that every one of India’s 140 million farm holdings adopts the technology.
If the Conclave had a technical centre of gravity, it was the session on Smart Agriculture: Technology, AI and Precision Farming, and the keynote that framed it.
Dr Jitendra Singh’s argument was straightforward and, on its own terms, correct. Climate change is among the biggest challenges facing global agriculture. Farmers facing it need better decisions, and better decisions need better information. Satellite technology, weather forecasting systems, resource mapping, drone-based surveys and real-time advisories now let farmers make informed choices on sowing, irrigation and crop management. Advanced forecasting in particular can help a farmer anticipate monsoon variability and select crops accordingly, reducing losses under shifting conditions.
Behind the advisory layer sits research. The minister pointed to work across departments under the Ministry of Science and Technology on climate-resilient crop development, genomics, crop improvement, pest-resistant varieties, precision farming and resource optimisation. Artificial intelligence, he said, is becoming “an indispensable tool” for predictive crop management, precision irrigation, weather-based advisories and the efficient use of agricultural inputs.
And agriculture is not only the patient. Invoking India’s net-zero-by-2070 commitment, he placed farming inside the clean-energy transition — biofuels, renewable energy, and the conversion of agricultural and food waste into value, citing the programme to turn used cooking oil into biofuel. Waste becomes feedstock; the sector that suffers climate change also helps mitigate it. It is a genuinely useful reframing, and it is not rhetoric: the circular-economy logic is the same one driving India’s ethanol blending programme.
The number, and where it comes from
The line that travelled furthest was this: AI-driven optimisation alone can help each farmer save nearly ₹5,000 annually, creating an estimated ₹70,000 crore in value addition for the overall agricultural economy.
It is worth doing the arithmetic, because the arithmetic is the argument. ₹70,000 crore divided by ₹5,000 gives 14 crore — 140 million. That is, almost exactly, the number of agricultural holdings in India.
This is not a journalist’s reconstruction. The minister supplied the basis himself, at a different event. Addressing the AI4Agri 2026 Global Conference in Mumbai on 22 February — a Government of Maharashtra summit held as a satellite of the India AI Impact Summit — he put it explicitly: India’s 140 million farm holdings, most of them small and marginal, “could together generate an estimated ₹70,000 crore in annual value if AI-enabled advisories help each farmer save even ₹5,000 a year through better input timing, pest prediction and market linkage.”
So the figure is internally consistent. It is also, read carefully, a ceiling rather than a forecast. ₹70,000 crore is what India’s farm economy would gain if every single holding in the country — including the smallest and most marginal, including those without smartphones, reliable connectivity, or literacy in the advisory’s language — adopted AI-enabled advisory and each realised the full ₹5,000. It is a total addressable value, of the kind a strategy consultant would put on slide three. It is not a projection of what will happen.
At the Delhi Conclave, the ₹70,000 crore was cited without that denominator, and — as at least one technology outlet noted at the time — without a named source for the underlying estimate. Neither omission is dishonest. Both change what the number means. A figure that describes universal adoption should be reported as describing universal adoption, particularly by a sector whose central historical problem is precisely that useful things do not reach everybody.
What actually stands behind the rhetoric
The Conclave, being a conclave, produced speeches. It is worth setting them against the machinery that exists, because the machinery is more impressive than the rhetoric and gets less attention.
The Union Budget 2026–27 has proposed Bharat-VISTAAR, a multilingual AI advisory tool integrating AgriStack portals and ICAR’s package of practices with AI systems, intended to deliver customised advice and reduce farm risk. BharatGen, India’s government-owned large language model ecosystem, has produced Agri Param — a domain-specific agriculture model operating in 22 Indian languages, launched by the Prime Minister. The ₹10,372-crore India AI Mission is building sovereign compute, datasets and start-up infrastructure. The Department of Science and Technology is backing an open, interoperable India AI Open Stack so that agri-AI built anywhere in the country can plug into a national framework. Maharashtra’s MahaAgri-AI Policy 2025–29 carries a ₹500-crore implementation allocation. A National Agri-AI Research Network has been proposed, linking DST, state governments, ICRISAT, ICAR and international institutions to build India-specific foundational datasets on crops, soil and climate.
The design philosophy on display there is notably sober: small, purpose-built models trained on Indian soil types, climate zones and crop varieties, deployable over low connectivity through mobile phones and farm equipment. Drone and satellite mapping is already feeding verified land and soil data into Soil Health Cards and the Svamitva Mission. This is not vapour. It is a state building agricultural digital public infrastructure at considerable scale.
None of that was announced at the Conclave. Which is rather the point: the Conclave’s technology agenda is downstream of decisions already taken in Budget documents and mission architectures. Its function was to socialise them among the people who will build on top.
The diffusion problem
The most important thing the minister said about technology was not about technology. It was that awareness and communication must keep pace with innovation — that digital communication, short-form video, multilingual content and AI-powered language tools are needed so farmers and rural youth can actually access and act on scientific knowledge.
That is a concession, and a significant one. It relocates the bottleneck. India’s constraint in agricultural technology has rarely been invention; the ICAR system, the CSIR laboratories and the state agricultural universities have produced varieties, practices and advisories for decades. The constraint is extension — the last-mile institutional machinery that carries a recommendation from a research station to a five-acre holding in Vidarbha and persuades the person farming it to change what they do. That machinery has been under-resourced for a generation.
The AI-language answer to this is real. A model that gives advice in Marathi, Bhojpuri or Kannada removes a genuine barrier. But language is one barrier among several, and not the tallest. Trust, credit, tenure, input availability, market access and the simple fact that a marginal farmer cannot afford to be the one who experiments — none of these is solved by better translation.
There is a further asymmetry rarely stated. An advisory that is right nine times out of ten is an excellent product for a corporate farm and a potentially ruinous one for a household with a single season’s buffer. The farmer who follows a confident recommendation and loses a crop bears the entire cost; the model bears none. Until advisory systems carry some liability, or are paired with the insurance and credit instruments that let a smallholder absorb being wrong, adoption will remain rational to refuse — and refusal will be misread as backwardness rather than as arithmetic.
The honest position
The technology case made at Delhi is correct in direction and unproven in magnitude. Climate stress is real, decision-support genuinely helps, satellites and forecasting do reduce losses, and India is building the digital plumbing to deliver them at national scale. That is a serious agenda and deserves to be taken seriously.
The ₹70,000 crore, though, is doing rhetorical work its arithmetic cannot support. It describes a world in which all 140 million holdings adopt. The measure of this agenda a year from now will not be the estimate. It will be the adoption rate — how many farmers actually used a Bharat-VISTAAR advisory, in what language, and whether it changed a decision that changed an outcome. That number does not yet exist. It is the only one that matters.
Sudhakar Bhima
