If Farming Keeps Getting Smarter, Why Does Food Keep Getting Dearer?

Every few months the question arrives in some form, usually with an edge to it. We have drones over the fields and satellites reading soil moisture. We have improved seed, precision irrigation and an entire industry of farm software. So why is the vegetable basket dearer than it was last year, and the year before that?

The question contains an assumption worth pulling out, because almost everything else follows from it: that the price of food is set on the farm. It is not. Technology has done a great deal to what happens between sowing and harvest. The consumer price is decided almost entirely by what happens afterwards — and by what the farmer had to spend before.

Here are the five places the gain leaks away.

The Reserve Bank's work on farm-to-retail margins found that of what a consumer pays for tomatoes, roughly a third reaches the grower. Onions return about 36% to the farmer and potatoes about 37%. The rest is distributed across commission agents, wholesalers, transporters and retailers — a chain that a better seed variety cannot shorten.

The instructive comparison sits in the same research. Dairy farmers capture something closer to 70% of the retail price, and the reason has nothing to do with cows being more technologically advanced than tomatoes. It is that milk moves through an organised, largely cooperative chain with fewer hands in it. The gap between 33% and 70% is a measure of market structure, not of agronomy.

A study commissioned by the Ministry of Food Processing Industries and carried out by NABCONS across 54 commodities put India's post-harvest losses at about ₹1.53 lakh crore a year — roughly 2.35% of GDP. Fruit losses run between 6% and 15%; vegetables between about 5% and 12%. Guava loses 15.05% of its crop after harvest; tomato, 11.61%.

Read that against a yield improvement and the arithmetic turns uncomfortable. A technology that lifts output by a tenth, on a crop that loses a tenth on the way to market, has fed nobody and lowered no price. It has simply increased the volume passing through a chain that was already the constraint. The bottleneck moved. Most of our investment did not move with it.

There is a quiet category error in the original question. We speak of technology as though it arrives free and its only effect is to raise output. On a farm, it arrives as an invoice: certified seed, agrochemicals, diesel, hired machinery, drip sets, and the electricity to run any of it.

Between 2013 and 2025 the weighted average cost of cultivation, as computed by the Commission for Agricultural Costs and Prices, rose by something in the range of 78% to 90%. Yields did not rise by anything close to that. A farmer adopting more technology is frequently running faster to hold position — and the cost of the inputs is eventually in the price of the output.

Small and marginal holdings — under two hectares — make up 86% of India's farms and work about 47% of the operated area. The average holding was 1.08 hectares at the last full Agriculture Census, and a marginal farmer's typical plot is around 0.38 hectares.

Capital equipment does not amortise on 0.38 hectares. That is not a failure of the farmer's ambition; it is arithmetic. Meanwhile the roughly $2.4 billion that has gone into Indian agritech since 2014 has largely funded software — marketplaces, advisory apps, lending, traceability. Useful work, much of it. But it has mostly been built for the chain, and the farm gate still opens onto the same road it always did.

In June and July 2026, onion prices climbed nearly 25% on weather-driven supply disruption. Tomato has swung by more than 60% inside a single year. No amount of average yield improvement protects a consumer from a bad fortnight in a growing region, because perishables cannot be stored through the shock and cannot be moved quickly enough to replace what was lost.

This is the part most easily missed. Technology has raised the mean. It has done very much less to narrow the variance — and the variance is what a household actually experiences at the vegetable cart. An average that improves while the spikes stay just as sharp does not feel like progress to the person buying dinner.

There is one more reason food costs more, and it deserves to be stated plainly rather than filed under problems. A significant share of the increase in the cost of cultivation is rising rural wages.

Farm labour is getting paid more than it was. In an economy where every other sector's wages are climbing, agriculture cannot hold its costs down without holding its workers down — and food produced by people who are paid properly is going to cost more than food produced by people who are not. That is not an inefficiency to be engineered away. It is the single most defensible line in the entire price build-up.

Which leaves a sharper question than the one we started with. If we want food that is both affordable and fairly produced, the room to manoeuvre is not in the field, where the marginal gains are getting harder and more expensive to win. It is in the two-thirds of the rupee that never reaches the farmer: in cold chains that stop the losses, in grading and aggregation that let a small farm sell like a large one, and in removing handoffs that add margin without adding value.

The technology worked. We pointed most of it at the half of the problem that was already working.