Autonomous Farming: Working Machines, Failing Balance Sheets

Two large harvesting machines working side by side across a green field

Autonomous weeding and spraying equipment works. That is no longer the interesting question. The interesting question is whether anyone can build a business selling it, and between 2024 and 2026 three companies answered no: FarmWise shut down on 1 April 2025, Small Robot Company entered administration in February 2024, and Monarch Tractor’s remaining assets went to Caterpillar in April 2026 after it had raised more than $240 million. The surviving machines are real, named and shipping. So is the capital problem underneath them.

See and Spray, in John Deere’s Own Figures

John Deere’s targeted sprayer is the most widely deployed AI tool in row-crop agriculture, and Deere publishes the numbers itself. The company states that See and Spray customers achieved 59 percent average herbicide savings in the 2024 season across more than one million acres, an estimated 8 million gallons of mix. For 2025, in an account by Ryan Daily published by AgNavigator on 10 November 2025, Deere says the system covered 5 million acres and saved 31 million gallons.

In that same piece Deere claims an average yield gain of 2 bushels per acre, rising to 4.8 bushels for some growers, attributed to third-party researchers and universities with no institution, author or method identified. Joshua Ladd, a Deere marketing manager, is the only named source in the article. The herbicide savings are a supplier’s measurement of its own product; the yield figure does not yet have enough behind it to be treated as established.

The One Independent Cost Comparison Worth Reading

The best economic data in the sector comes from a Western Growers study issued in March 2024 and reported by Tim Hearden of Western Farm Press on 21 February 2025. It set laser weeding against hand weeding on real acreage rather than in a brochure.

The machine in question is Carbon Robotics’ LaserWeeder, which kills weeds with lasers rather than chemicals or blades. It is real, shipping and named, which already separates it from most of what gets written about robot farming.

The study’s figures are set out below. They favour the machine and still explain why adoption is hard: a saving on that scale exists only if the buyer can finance the equipment and then keep it running long enough to spread the cost, which the arithmetic assumes at 18 hours a day.

  • Braga Fresh, growing organic baby leaf crops in California’s Salinas Valley, reported hand weeding at $900 per acre and 90 minutes per acre, using three crews of 25 employees.
  • The same operation put laser weeding at $267.72 an acre, on a $1.2 million machine depreciated over five years and run 18 hours a day.
  • Two machines covered 4,700 acres at that operation.
  • Pictsweet Farms in Tennessee reported comparable savings and dropped its chopping crews.
  • Machine capability is cited at over 200,000 weeds an hour, up to 99 percent lethal.

Four Dollars an Acre Is the Whole Constraint

Andrew Rosenblum’s piece for Ambrook Offrange, published 27 April 2026, carries the decisive number. Ankit Chandra, a researcher at the University of Nebraska-Lincoln, puts what row-crop farmers can afford for new technology at roughly $4 an acre, and says bluntly that the farmer does not have that amount to invest in new technologies.

Throughput is the other half of the squeeze. FarmWise’s Vulcan weeded about 100 acres a week using two machines, and Andy Jobman, a Nebraska corn and soybean farmer, called weeders working at around 50 acres a week just not necessarily scalable for row crops running past 1,000 acres. Specialty crops with high hand-labour bills can absorb the price. Corn and soybeans cannot.

The investment climate moved the same way. Agtech venture funding fell from $11 billion in 2021 to $6 billion in 2024. Jenny Lemieux, chief executive of Vivid Machines, told Ambrook that farmers value demonstrable savings on materials or labour over equipment features.

Three Companies That Ran Out of Road

FarmWise Labs of San Francisco closed on 1 April 2025. Chief executive Tjarko Leifer cited an inability to raise further venture capital, low commodity prices, tariff-driven component costs, and legacy machinery brands down 40 percent year on year. More than a dozen Vulcans were in specialty crop fields at closure; customers were told the machines keep working without software updates but work better with them. Ambrook reports the Vulcan carried a $645,000 price tag and that FarmWise raised $65 million. The salad producer Taylor Farms later bought the assets.

Small Robot Company of Salisbury, England, founded in 2017 by Ben Scott-Robinson, Sarra Mander and Sam Watson Jones, had grown past 50 employees. It entered administration on 1 February 2024 with Kroll appointed, announced the following day, after a signed term sheet failed to convert into investment. It had raised $13.18 million across six rounds, the last in September 2022, and had built Tom for monitoring, Dick for weed zapping, which never got past early trials, Harry for planting and Wilma for weed identification. It said it was a victim of the valley of death and that the United Kingdom lacks the hardware funding ecosystem available in the United States.

Monarch Tractor of Livermore, California, founded in 2018 by Praveen Penmetsa, Mark Schwager, Carlo Mondavi and Zachary Omohundro, sold its MK-V electric driver-optional tractor at up to $100,000 and raised more than $240 million, including a $133 million Series C in August 2024. Foxconn ceased MK-V production in August 2025 after selling the Lordstown facility. A WARN Act notice followed in November 2025; sources put the headcount at 89 and at 102, and this article prints both. Equipment was auctioned in February 2026, and in April 2026 the remaining assets went to Caterpillar.

Precision Farming Dealer reports Monarch’s statement of 10 April 2026, posted on LinkedIn without individual attribution, saying its core technology had been acquired by a large global equipment manufacturer and conceding that building and scaling a new tractor platform came with unforeseen challenges. The same coverage records a November 2025 lawsuit from the Idaho dealership Burks Tractor alleging breach of contract over defective autonomous tractors.

Naio’s Smaller Machines Took a Different Route

Not every builder chased the large machine. Naio Technologies, founded in Toulouse in 2011 by the robotics engineers Gaetan Severac and Aymeric Barthes, builds Oz for market gardeners, Dino for vegetables and Ted for weeding between vine rows, and closed a Series A of 14 million euros, about $15.53 million, in January 2020. The Robot Report described nearly 150 robots in circulation worldwide, more than 20 of them Teds, with French trials involving Jas Hennessy and Company and Bernard Magrez Grands Vignobles.

Small, crop-specific and priced far below a $1.2 million laser rig, that approach has so far outlasted several better-funded rivals. It is not a general solution to weeding a thousand acres of soybeans, and it does not claim to be. The pattern across the survivors is narrower ambition and equipment sold into crops where hand labour is expensive enough to justify it.

Sources: John Deere · Farm Progress · Ambrook Offrange · Farm Progress · The Robot Report · Precision Farming Dealer

More articles to read