Weeding by Laser, Without a Drop of Herbicide

In a lettuce field in California's Salinas Valley, a machine as wide as a semi trailer crawls forward after dark. Beneath it, some thirty infrared beams flicker in pulses invisible to the eye. Each one finds an unwanted seedling, heats it for a fraction of a second and bursts it at the cellular level. No chemical, no blade, no hand. By morning the rows are clean and the soil has not been turned.

The scene is no longer experimental. The LaserWeeder, built by the American company Carbon Robotics, has already weeded more than 250,000 acres and destroyed over fifteen billion weeds across a hundred crops, according to its maker. More than two hundred of these machines now work on farms in North America, Europe and Australia. Behind the feat lies a simple promise: doing without both chemical herbicides and the hand-weeding crews that are growing scarce. The question is what it costs, and who holds the controls.

Two things that vanish from the field

To weigh what is happening, look at what the beam replaces. In vegetable farming, labor accounts for up to 40 percent of production costs, sometimes half. Hand weeding is both expensive and hard to staff: over 160 dollars an acre for organic lettuce, up to 440 dollars for organic spinach, more still in dense baby greens. When crews stop showing up, some growers shrink their acreage, others consider leaving the business.

The second absentee is the herbicide can. The laser sprays nothing and leaves no residue on the leaf or in the soil. As weeds grow ever more resistant to glyphosate, the chemical route is itself losing reliability. A 2024 study by the Western Growers Association put a figure on the shift: laser weeding cuts the cost of the task by roughly 40 percent.

What the grower gets back, in the end, is control over a job that depended on an uncertain workforce and a fading product. For anyone who grows what we eat, that is not a management detail; it is the ability to carry a season without relying on what may fail to turn up. The machine works at night, never tires and asks for no visa, three things a hand crew cannot promise.

Precision instead of the can

The principle is a very fast loop. Onboard cameras film the row, a vision model tells crop from weed to the centimeter, then aims a beam at the weed's growth point. The targeted plant cooks in milliseconds, with no tilling of the soil. It is the opposite of mechanical hoeing, which turns the earth, brings fresh seeds to the surface and dries out the top layer.

That precision has knock-on effects. Soil left undisturbed keeps its moisture and microbial life; produce that has met no herbicide reassures a growing share of shoppers. Each shot is logged as well, so the machine quietly maps where weeds press hardest, turning a chore into data the farm can act on the next season. Other makers take a middle road: precision sprayers from Bilberry, Trimble or Naïo wet only the plants they detect and cut herbicide use by up to 90 percent. The laser aims further, at removing the product altogether, at least in the rows where it works.

The price of autonomy

This freedom carries a steep tag. A LaserWeeder sells for more than a million dollars: the large G2 600 model tops 1.4 million, before annual support fees. Carbon Robotics designed its G2 range for farms of 80 to 800 acres, with a payback period of about three years. For a small grower the math stays out of reach; the machine is aimed first at large specialty farms, the ones already running on tight margins and scarce labor.

More to the point, autonomy moves rather than settles. The grower stops depending on chemical suppliers and labor brokers, but binds himself to a single manufacturer, its proprietary software and its yearly updates. A hoe belonged to whoever held it; the beam answers to a company, its software licenses and its goodwill. If the firm stumbles or changes its terms, the field stops weeding itself. It is a trade, not a release.

Where the laser stalls

The method has blind spots. It works best on small seedlings: a weed that has grown too large demands more energy and slows the machine, forcing several passes through the season. The pace stays modest, around two acres an hour for the early models; a 1,500-acre farm ties the machine up for some ten days straight. Cold and wet foliage, which sap the heat, trim the effectiveness further.

Finally, the laser remains a row-crop affair: lettuce, carrots, onions, spinach, broccoli, tomatoes. It has nothing to offer the broadacre wheat and corn fields where most herbicide is actually sprayed. The promise of a chemical-free field therefore holds for one slice of farming, the most labor-intensive, not for the whole landscape.

The movement is widening all the same. France's Naïo has announced a laser weeding head for 2027, sensors keep getting cheaper, and per-acre rental plans are starting to open the door to mid-sized farms. The real question is probably not the beam but the controls: as our vegetables grow under the eye of algorithms, one set of dependencies gives way to another, quieter and more concentrated. Who holds those controls will decide what this recovered freedom is really worth.