The future of weed control. Food production's most intractable challenge & the Technologies re-shaping the $35 billion herbicide business
Weed control is becoming a very different business. Herbicide resistance, labor constraints and pressure to reduce chemical use are pushing growers beyond a single product or technology and toward combinations of chemistry, biology, precision application, robotics and physical control.
Weeding out the weeds isn’t just one of agriculture’s oldest problems, it’s also the hardest to remedy using new technology, while resistance to existing solutions is making weed control one of food productions most expensive to resolve. Research published in Weed Technology estimates that without weed control, losses across seven major crops in the United States and Canada could reach roughly US$51 billion a year, with corn alone accounting for US$26.7 billion. That economic importance is reflected in a global herbicides market worth about $35.4 billion in 2025.
Herbicide resistance is reducing the number of reliable options growers can rotate between. The International Herbicide-Resistant Weed Database now records 548 unique cases of resistance across 275 weed species, 102 crops and 76 countries, with resistance documented to 21 of the 31 known herbicide sites of action. At the same time, hand-weeding labor is harder to find, regulators continue to remove active ingredients, and food companies are asking more questions about residues and chemical use.
The combination of those factors is a perfect storm and changing how innovation is happening. Chemistry, biology, precision application, robotics and physical control are increasingly being developed as complementary parts of weed-management programs. As this article will demonstrate the best option today will be the adoption of novel bioherbicides, e.g. Harpe, together with precision application technologies such as see and spray, combined with existing chemistry where allowed, or alone where products such as Roundup are no longer allowed.
Five factors are beginning to reshape the business.
New chemistry: a long innovation gap starts to close
For roughly three decades, very few herbicides with genuinely new modes of action reached the market. As resistance accumulated to existing chemistry, the need for new molecular targets became increasingly urgent. That long innovation gap is now beginning to close, helped by new approaches to target discovery, computational screening and AI-assisted molecule design.
BASF was among the first to bring genuinely new chemistry back to market with cinmethylin-based Luximax, a pre-emergence Group 30 herbicide that disrupts cell-membrane processes during germination and has shown control of multiple-resistant ryegrass. FMC followed with tetflupyrolimet, the first Group 28 herbicide, now registered in Peru for pre- and early post-emergence control of difficult grass weeds in rice. Bayer is developing icafolin-methyl, a new post-emergence herbicide for broadacre crops, using its CropKey platform to combine target-based molecule design, formulation and AI, with the herbicide intended for lower-dose and targeted application alongside existing chemistry such as glyphosate.
The discovery process is changing too. Moa Technology says it has screened more than 900,000 compounds and identified more than 80 potential novel mode-of-action areas through its discovery platforms; it is also developing “Amplifiers,” molecules that are not herbicidal themselves but may reduce the amount of another herbicide required. Enko borrows technologies from pharmaceutical research, combining DNA-encoded libraries, structure-based design, AI and machine learning to screen enormous molecular libraries against selected biological targets. These platforms matter because they shift herbicide discovery from largely testing compounds and seeing what works toward identifying targets first and searching much larger chemical spaces for molecules designed to act on them.
Penn State’s 2026 weed-control update is a useful reality check: most of what reaches growers each season is still a new mixture, formulation or use of existing chemistry, rather than a genuinely new mechanism.
Bioherbicides: the biggest gap in biologicals
Biologicals have made real progress in fungicides and insecticides, but weed control has been much harder. CropLife has called bioherbicides “the biggest product gap” in the sector, and that is a fair description. Research in Frontiers in Agronomy and Weed Science shows that some natural compounds can disrupt membranes, microtubules and other processes involved in weed growth, although their exact modes of action are not always fully understood. The harder part has been turning that science into products that perform consistently in the field, make economic sense per acre and fit existing weed-control programs.
One company addressing that gap is Harpe Bioherbicide Solutions. Its technology uses plant-derived compounds from sources including mint and lemongrass, while its patent covers Mentha-based herbicidal compositions containing naturally occuring bioactive compounds, including synergistic combinations with conventional herbicides. That gives Harpe two interesting routes: using the biological chemistry on its own or integrating it with existing herbicide programs. Harpe has reported more than 1,000 greenhouse and field trials by 2024 and activity against more than 30 resistant weeds. Rather than positioning biological chemistry simply as a replacement for conventional herbicides, Today Harpe is the lead technology in non-selective contact herbicides for both pre-emergence and post-emergence burndown. in March 2026 they signed a development agreement with SBM Life Science to co-develop and commercialize natural weed-control products for the U.S. home and garden market. The company is also exploring how that chemistry fits precision application and paired its bioherbicide with Verdant Robotics’ AI-guided Sharpshooter to target individual weeds spray across the field.
An established commercial comparison is Beloukha, a sunflower-based pelargonic acid contact herbicide now sold by Albaugh. It provides fast, non-selective burndown of broadleaf and grass weeds but no residual weed control, and although less effective in field situations it illustrates both what plant-derived chemistry can already do and the performance gap newer bioherbicides are trying to close.
Cameras and AI: spraying the weed, not the field
Targeted spraying changes the unit of weed control from the field to the individual plant. Cameras scan the crop in real time, AI distinguishes weeds from crops, and individual nozzles fire only where treatment is needed. The potential to cut herbicide use is clear, although a 2026 review of 27 economic studies found that profitability still depends heavily on weed pressure, machine utilization and technology cost, with relatively little multi-year evidence from commercial farms.
John Deere has taken targeted spraying to the largest scale with See & Spray, built from the Blue River Technology acquisition. Deere reports that the system was used across more than five million acres in 2025. A three-year trial by the University of Arkansas found post-emergence herbicide use in soybeans cut roughly in half, while an Iowa State demonstration across 415 acres recorded savings of 44 to 91%, depending on initial weed pressure. That range is the honest number: savings track weed density, not marketing. Verdant Robotics takes a more surgical approach with SharpShooter, using vision AI and aimable turrets to deliver micro-shots to individual plants. Verdant reports that a trial conducted by Rutgers New Jersey Agricultural Experiment Station and Cornell AgriTech matched broadcast weed control in spinach while avoiding crop stunting and producing higher yields, suggesting precision may add value by protecting the crop as well as saving chemistry.
The rest of the market is differentiating itself through how technology is deployed. Greeneye Technology retrofits existing sprayers with a dual-line system that broadcasts residual chemistry while spot-spraying non-residual herbicide, with Montana State reporting an average 87% reduction in non-residual use. Ecorobotix uses plant-by-plant AI with its ARA sprayer, treating areas as small as roughly 6 by 6 cm, while ONE SMART SPRAY combines Bosch camera hardware with BASF agronomic software and has pursued OEM integration with major equipment manufacturers. Bavaria’s Allgäu Automation is a more specialized case: its RumboJet targets individual weeds in grassland, with around 130 machines operating across six European countries by late 2025.
There is a business tension here. Precision spraying means selling fewer liters, but potentially more value per acre through higher-value chemistry, biologicals and smarter application. The opportunity may shift from selling product volume to selling intelligence around each application.
Robots and mechanical weeding: machines that physically remove weeds
Mechanical weeding never disappeared. Machine vision is what has made it precise. A 2026 review in Sensors found robotic intra-row cultivators removed 18 to 41 per cent more weeds and cut subsequent hand-weeding by 20 to 45 per cent compared with standard cultivation. The trade-off remains speed: as machines travel faster, accuracy falls and crop injury increases.
Greenfield Robotics is pushing mechanical robotics into broadacre crops with BOTONY, using RTK navigation, cameras and autonomous fleet software to cut weeds mechanically in crops including soybeans, milo, cotton and corn; a 2026 SEC filing reported orders for 70 robots, with 60 shipped by early June. FarmWise shows the harder side of the business: its AI-guided Vulcan weeder worked, but the company wound down in April 2025 before being acquired by Taylor Farms, which kept support for customers. Working technology and a working business model remain two different challenges.
Naïo had more than 350 robots deployed across five continents by late 2025 and now has two particularly relevant platforms: OZ, a small autonomous assistant for seeding and mechanical weeding in specialty crops, and TED, an autonomous straddler for mechanical under-vine weeding. Stout Industrial Technology takes another route with a tractor-drawn Smart Cultivator that uses machine vision and AI to control mechanical blades, while its Cortex platform turns the same plant-level vision data into agronomic analytics.
Lasers, electricity and weed seed destruction
The most radical approach removes chemistry altogether. A 2025 peer-reviewed study by Rutgers and Cornell researchers compared a commercial laser weeder with conventional herbicides in beet, spinach and pea. Laser treatment matched or exceeded herbicide control for several important annual weeds, reduced end-of-season weed biomass by at least 97 per cent compared with untreated plots, and caused almost no crop stunting.
Carbon Robotics has taken laser weeding from specialty vegetables toward broadacre farming, with more than 150 machines operating on over 100 farms in 14 countries and a new 40-foot LaserWeeder aimed at organic corn and soybeans. It uses computer vision, AI and lasers to destroy individual weeds, but the economics are still demanding at about US$1.6 million and 5 to 10 acres per hour. Electrical systems offer a different route: LASCO’s Lightning Weeder uses electrodes above the crop canopy to kill taller escapes such as Palmer amaranth and waterhemp, while RootWave uses high-frequency current to destroy weeds through the plant and roots, bringing modern electronics to a weed-control idea that has been around for decades.
Physical control can also happen at harvest. Impact mills such as the Redekop Seed Control Unit, Seed Terminator and integrated Harrington Seed Destructor pulverize weed seed in the chaff before it returns to the field. Virginia Tech trials found kill rates above 99% for problematic soybean weed seeds and above 89% in wheat when the seed reached the mill. The trade-off is additional horsepower and fuel use, but every viable seed destroyed at harvest reduces the pressure carried into the next season.
The new economics of weed control
Different crops, weed populations, labor costs and margins will favor different combinations of technology. In practice, that may mean residual chemistry followed by targeted spraying, biologicals where they fit, mechanical or physical control where the economics work, and greater attention to preventing surviving weeds from replenishing the seedbank.
For agribusiness, the larger shift is from weed control sold by the liter to weed control managed by the acre. Chemistry, biology, data and machinery are converging into the same system. As stated at the outset the most viable alternative to today’s realities is the use of bioherbicides together with precision application technologies. The future of sustainable and profitable crop production will belong to those who can make that combination work reliably, economically and on scale.
Thanks to Camila Ulloa for her help in researching and writing this and several industry insiders who offered edits and additions.
