The Tractor Plows on Its Own, but Who Holds the Keys?

In the spring of 2026, in an Iowa field, a farmer climbs down from his tractor cab, takes out his phone, slides a control across the screen and walks away. The machine, an 8R weighing more than fifteen tonnes, starts on its own, opens its first furrow and will carry on without him to the far edge of the plot. John Deere has a name for the gesture: "swipe to farm," a flick of the screen that sets the tilling in motion.

The manufacturer is field-testing fully autonomous tractors across eighteen states and preparing a nationwide rollout. The stakes are anything but abstract. American farms struggle to hire, seasonal labour is drying up, and the windows for planting or harvest are sometimes counted in days. Handing the wheel to software promises to give the farmer back the hours he once spent, alone, pacing the field at the speed of an engine.

What the machine can do with no one aboard

Autonomy does not appear from nowhere. The 8R relies on a cluster of cameras covering a full three hundred and sixty degrees and on vision models trained to spot an obstacle, a ditch, a human shape, in thick dust. The system pinpoints the machine to the centimetre, plots its path and corrects it without pause. Deere inherited twenty years of satellite guidance: the novelty is not following a straight line, it is deciding on its own when to stop.

From his phone, the operator tracks the tractor's position, its engine load, its fuel use, the ground already worked. An alert warns him whenever the machine meets something it cannot interpret; he then studies the image, makes the call, sets it going again. One operator can thus watch several machines at once, or attend to something else while the field tills itself.

This is hardly the first robot to enter a farm. Arms already pick strawberries, drones spray row by row, sensors read the moisture in the soil. But the tractor remains the centrepiece, the tool around which the whole operation is built. To see it advance with an empty cab changes the very nature of the job.

Time given back, the body spared

The clearest benefit is measured in hours. A spring ploughing, an autumn tillage, these mean whole days spent fixed on a furrow, nudging the same path a quarter-turn at a time. A task demanding enough to forbid any distraction, repetitive enough to wear attention thin. Making it autonomous frees the farmer from labour that taxed his body without calling on his judgement.

The gain goes beyond fatigue. The machine does not sleep. It can work at night, seize a weather window of a few hours, run plot after plot when time is short and every forecast shower threatens the harvest. Where a farmer once had to choose between sleeping and finishing, the software finishes while he sleeps. For a farm short of hands, that is no luxury; it is sometimes the difference between a crop brought in and a crop lost.

There is, finally, a quieter kind of ease. Watching from the kitchen rather than the seat, keeping an eye on the screen while settling a bill or greeting a visitor: autonomy does not abolish the work, it peels it away from the engine. The farmer stops being an extension of his machine and becomes its conductor again.

The empty cab, the binding contract

Then comes the catch, and it is a large one. A tractor that drives itself is, first of all, a tractor that depends on software, and therefore on whoever owns it. Deere spent a decade locking access to its own machines. On 15 January 2026, the U.S. Federal Trade Commission, joined by Illinois and Minnesota, sued the manufacturer for restricting the repair of its equipment, forcing farmers to go through authorised dealers.

The figures cited by right-to-repair advocates give the measure of the grievance: three billion dollars lost each year to equipment downtime, one and a fifth billion in excess repair costs, for want of being able to step in oneself. In the middle of a planting window, a breakdown a neighbouring mechanic could fix in an hour can pin a machine for days while a certified technician makes his way over. In the spring of 2026, Deere agreed to pay ninety-nine million dollars and to open up, for ten years, its digital diagnostic tools. Repair groups remain unconvinced.

The more autonomous a tractor becomes, the heavier that dependence weighs. A machine steered by vision and satellite no longer rolls without its update, its connection, the maker's say-so. The farmer's promised autonomy rests, paradoxically, on a deepened reliance on a company he does not control, inside a software box he is not allowed to open.

Who owns the field that drives itself?

Beneath the mechanics lies a quieter question. The autonomous tractor does not merely plough: it records. Position, yield, soil condition, plot by plot, season after season. That data, which describes the farm better than its owner ever could, passes through the manufacturer's servers. Whoever holds it knows the operation in fine detail, and can spin it into services, advice, prices.

The farmer gains time and is spared real toil. But he trades an autonomy of action for a dependence on a system. He no longer holds the wheel; nor does he hold the code that replaces it, nor the keys to the vault where the encrypted portrait of his land sleeps. The machine works alone, and that is precisely why he is no longer quite the sole master of his own ground.

The driverless tractor, then, is not just one more convenience. At the scale of a single field, it raises the question every automation eventually raises: what are we willing to surrender in order not to have to do? The answer, here, is not settled in the cab. It is settled in a contract, an update, a line of code the farmer signs without always being able to read it.