John Deere's See & Spray Hits Cotton, Irrigation Startups Break Out, Drone Mappers Win

On a corn operation outside Hastings, Nebraska, a John Deere self-propelled sprayer crosses a May morning with 36 camera-and-nozzle modules making hundreds of spray/no-spray decisions per second. When the machine's vision system detects a waterhemp plant hidden inside the corn canopy, it fires a targeted herbicide burst—and passes over every surrounding corn plant untouched. Across three commercial seasons, that system has delivered up to 77% reductions in herbicide use and per-acre chemical bill cuts of $17–$38 depending on weed pressure and local input prices.

That machine—See & Spray Ultimate, the commercial product built on technology Deere acquired from Blue River Technology for $305 million in September 2017—now operates in corn, soybean, and (since 2024) cotton fields across the U.S. It is one of three converging forces reshaping American agriculture's AI infrastructure as of early 2026. The other two: precision-irrigation platforms cutting water consumption by 18–35% in drought-stressed growing regions, and a drone-imagery sector that has moved from a scouting novelty to an underwriting input that crop insurers are beginning to price into policies.

See & Spray's Architecture—and What the Numbers Actually Show

The gap between Blue River Technology's 2017 acquisition and commercial deployment was longer than analysts expected, but the product that emerged was more capable than the prototype Deere bought. See & Spray Select (launched for the 2021 planting season) distinguishes corn plants from weeds in the inter-row space—meaningful but limited to between-row detection. See & Spray Ultimate (commercial launch 2023) runs full canopy detection and targets weeds within the crop row, the harder and more herbicide-intensive problem that had previously resisted automation.

The hardware architecture defines the product's edge. The Ultimate mounts 36 camera-and-nozzle modules across a 120-foot (36.6 m) boom. Each module runs edge inference on NVIDIA Jetson-class processing—Deere and NVIDIA demonstrated the integrated stack at CES 2022—making binary spray decisions at roughly 10 frames per second as the machine travels at 12–15 mph. The vision model was trained on tens of millions of labeled plant images from Deere's customer fleet across multiple growing seasons, a dataset moat that startup competitors cannot quickly replicate. As of the Ultimate's commercial launch, the system identifies more than 40 weed species, including waterhemp, Palmer amaranth, and giant ragweed—the three species that most commonly develop herbicide resistance in the U.S. Corn Belt.

Performance data: Deere's published claim is up to 77% reduction in herbicide use. University of Nebraska–Lincoln Extension trials across the 2022–2024 growing seasons documented 47–66% reductions in commercial soybean fields, with lighter weed pressure producing savings at the high end of that range. At herbicide costs of $35–$55 per acre for conventional blanket application, field savings run roughly $17–$38 per acre—enough to amortize the system's premium over a typical working life in Midwest operations above 1,000 acres. The EPA's ongoing Endangered Species Act consultation on glyphosate and dicamba has expanded Pesticide Use Limitation Areas across parts of the Midwest and Southeast; Deere is actively marketing See & Spray as a compliance tool, since its spatial application log can document reduced field-level herbicide load if growers are audited near PULA boundaries.

In 2024, Deere expanded the supported crop list to include cotton—a strategically significant addition. Cotton is among the most herbicide-intensive major U.S. crops, with glyphosate and dicamba costs regularly exceeding $45 per acre in the Delta and Southeast. Early field reports from Georgia and Mississippi operations indicate herbicide reductions of 50–65% in cotton, though large-scale commercial data is still accumulating as of Q1 2026.

The Precision-Irrigation Startup Race

Water is agriculture AI's second major input battleground. In California, where almond growers paid $150–$300 per acre-foot for surface water in drought years, and in the High Plains, where aquifer depletion threatens the long-term viability of irrigated corn and wheat, the commercial case for AI-driven irrigation scheduling has become financially undeniable.

CropX: Soil Intelligence at Scale

Tel Aviv-founded CropX runs a soil intelligence platform—wireless sensors measuring moisture, temperature, and electrical conductivity at multiple depths, paired with cloud ML that generates irrigation and fertilization recommendations. Following a $30 million Series B in April 2021, the company acquired Dutch startup Farm21 in 2022, adding proprietary IoT hardware to a previously software-centric stack. CropX reports deployments across 40+ countries as of 2025. Third-party validation by the USDA Agricultural Research Service in Colorado found water use efficiency improvements of 18–31% on corn-under-pivot systems, with the variation driven by baseline soil management quality at the time of enrollment.

Lumo: Robotic Gates in Wine Country

California-based Lumo targets wine grape and specialty crop growers with robotic valve actuators that automate existing drip and surface irrigation systems without replacing them. A $20 million Series B in 2022 led by DCVC (Data Collective) funded U.S. expansion. Lumo's field data from California shows 22–35% reductions in applied water compared to hand-managed schedules, with growers citing per-vine-block differentiation—impossible with manual valve management—as the primary agronomic benefit. The company expanded into Washington State's Columbia Valley in 2024.

Lindsay FieldNET Advisor: AI on Existing Pivots

For the hundreds of thousands of center-pivot systems already in the ground across the Great Plains, Lindsay Corporation (NYSE: LNN) has layered AI scheduling onto its Valley brand hardware through FieldNET Advisor. The platform integrates weather forecasts, evapotranspiration models, and field sensors to make run/don't-run recommendations and set application depths automatically. A 2023–2024 cooperative study with University of Nebraska–Lincoln Extension found Advisor-managed pivots applied 11–18% less water with no yield penalty in corn; in the 2023 drought year, Advisor fields yielded 3.4 bu/acre higher than hand-scheduled controls, attributed to better timing around peak crop water demand events.

Drone Imagery and Application: Who's Winning

The drone-ag sector split early into two distinct markets—application (spraying inputs) and analytics (generating data)—with different competitive dynamics in each.

Application: DJI Agras and the Operator Layer

DJI's Agras T40 and T50 remain the dominant application platforms in the U.S. market. The T40 (40-liter tank, 24 nozzles, released 2022) covers approximately 21 acres per hour at standard application rates—versus roughly 2–3 acres per hour for backpack sprayers. Iowa-based Rantizo, one of the first U.S. operators to receive an FAA BVLOS waiver for agricultural drone spraying (2021), has scaled to multi-state operations in Iowa, Illinois, and Ohio, offering contract application at $8–$14 per acre. Houston-based Hylio operates a comparable service in the South and Mid-Atlantic. Both companies represent the “spraying as a service” layer above the drone hardware—the model best positioned to capture value as BVLOS rules loosen and per-pilot productivity improves.

Analytics: Aerobotics, Taranis/Corteva, and Sentera

Aerobotics, the Cape Town–founded tree-crop specialist, has emerged as the clearest analytics winner in orchards and nut crops. Its AI models—trained on imagery from South African citrus, California almonds, and Australian macadamia—detect individual tree stress, canopy gaps, and pest signatures including false codling moth damage and citrus stem borer entry holes, with per-tree detection accuracy above 90%. A $17 million Series B led by Kleiner Perkins (2022) funded U.S. expansion; California almond growers are now the largest customer segment by acreage, using Aerobotics data to prioritize scouting routes and justify block-level spray decisions that can save $30–$80 per acre in unnecessary fungicide or insecticide applications.

In row crops, Taranis—the Israeli sub-millimeter aerial imaging company—was acquired by Corteva Agriscience in 2021 and folded into Corteva's Granular Farm Management Software platform. The integration gives Corteva's agronomist network drone-verified disease and pest maps linked directly to prescription outputs and seed/chemical recommendations. Minneapolis-based Sentera occupies a complementary segment: multi-spectral sensors compatible with DJI Mavic and Phantom platforms, turning a $3,000–$5,000 consumer drone into a stand count and NDVI mapping tool. Sentera's 2024 stand-count accuracy benchmarks—validated in third-party field trials—exceeded 95% agreement with manual counts in commercial corn fields, a metric that directly informs replant decisions worth $50–$120 per acre.

The Regulatory Picture: FAA, EPA, and the BVLOS Bottleneck

The single largest regulatory lever in agricultural drones is the FAA's Beyond Visual Line of Sight (BVLOS) framework. Under current Part 107 rules, most commercial operators must keep aircraft in unaided visual contact—a constraint that caps a single pilot to roughly 300–400 acres per day and makes large-scale contract application economically marginal at scale. The FAA published an NPRM for a performance-based BVLOS framework in 2024, explicitly citing unobstructed agricultural terrain as a prototypical low-risk environment for the new standard. Preliminary data from BVLOS research exemptions issued by the Northern Plains UAS Test Site in North Dakota—covering Red River Valley corn and soybean fields—suggests a single pilot managing three synchronized aircraft under BVLOS protocols can cover 1,200–1,500 acres per day, a 4× improvement over visual-line operations. If the rule is finalized, it restructures the unit economics of contract drone spraying more than any hardware improvement in the past five years.

Insurance: Slow but Directionally Clear

The USDA's Risk Management Agency administers roughly $160 billion in crop insurance liability annually. In 2023, RMA issued guidance allowing producers to use precision-ag yield data logged through certified farm management platforms—including John Deere Operations Center—as supplemental documentation for Actual Production History databases. The practical effect: growers with multi-year precision-ag records can establish higher, more defensible yield histories on their best-performing field polygons, reducing basis risk in indemnity calculations.

Several private insurers writing specialty crop policies in California and the Pacific Northwest have begun offering 3–7% premium reductions for operations using certified soil-moisture monitoring systems. These adjustments are modest but represent the first actuarial recognition that AI-managed irrigation statistically changes a farm's yield-loss risk profile. Broader adjustments are contingent on multi-year loss data that is still accumulating across the precision-ag installed base.

What's Still Missing

Three structural gaps slow adoption. First, connectivity: farms that would most benefit from real-time AI recommendations—dryland wheat in the Texas Panhandle, irrigated corn in rural Nebraska—still have unreliable broadband. Starlink has materially helped, but analytics pipelines that require cloud round-trips still fail intermittently in the field. Second, agronomic trust: adoption surveys consistently show operators over 55 trust county extension agents more than apps, and the extension system is underfunded and understaffed. Third, interoperability: Deere's Operations Center, CNH Industrial's AFS Connect, and AGCO's Fuse platform speak different data dialects, and third-party startups must maintain connectors to each. The industry's belated adoption of the ISOBUS and ADAPT data standards is resolving this slowly—but not at the pace the market demands.