Mining Automation and Greenhouse Automation: What’s Real

Reviewed August 2026 against the U.S. Mine Safety and Health Administration (MSHA), USDA’s National Agricultural Statistics Service (NASS), and GlobalData’s Mining Intelligence Center.

Try it: Run your own numbers →

On This Page

Mining automation is no longer a pilot program: GlobalData’s Mining Intelligence Center counted 3,832 autonomous haul trucks running at surface mines worldwide in July 2025, up 84% from 2,080 a year earlier, according to reporting by Mining Technology. Greenhouse automation news is quieter but just as measurable — USDA’s newest horticulture census puts U.S. sales of food crops grown under protected structures at $1.01 billion for 2024, up 44% since 2019, while a global market-research forecast puts the wider controlled-environment agriculture (CEA) sector at $24.72 billion in the United States alone for 2025. This article covers both stories on their own terms, with sourced figures, four charts built from that data, and a calculator for sizing an autonomous-haulage business case.

Mining Automation vs. Greenhouse Automation: The Numbers

These two automation stories share a page because Farmonaut works across both mining intelligence and agricultural monitoring, not because the underlying markets overlap. The table below lines up what’s actually documented for each, with the source and date attached to every figure so you can refresh it yourself.

Mining Automation: How Big Is It, Really?

GlobalData’s Mining Intelligence Center tracks autonomous, autonomous-ready and tele-remote haul trucks, drills and load-haul-dump machines across the world’s surface and underground mines. In 2020, fewer than 1% of that tracked fleet fell into those categories; by the time Mining Technology reported GlobalData’s figures on November 20, 2025, the share had risen to 4.2% of the total operating population of surface/underground trucks, surface drills and LHDs. Five countries account for more than 90% of that installed base: China (56%), Australia (21%), Canada (7%), Chile (5%) and Brazil (1.4%).

Unit counts tell the same story with more texture. GlobalData’s separate truck-specific tracker recorded 2,080 autonomous haul trucks in operation in July 2024, rising to 3,832 by July 2025 — an 84% year-over-year increase, with the total projected to approach 5,000 by the end of 2030. China’s fleet alone grew from 562 to 2,090 trucks over that span; Australia’s grew more slowly, from 927 to 1,024. Canada (344) and Chile (208) round out the top four as of July 2025.

GlobalData’s tracker updates continuously as new sites go autonomous or expand existing fleets. Before you cite the 3,832 figure in a proposal, check Mining Technology’s coverage or GlobalData’s Mining Intelligence Center directly for whatever count is live when you read this.

Autonomous Haulage and Drilling Fleets

Autonomous haulage remains the anchor of mining automation because haul trucks are the highest-headcount, highest-hazard machines on a surface mine, and because retrofitting an existing truck costs less than replacing an entire fleet. Caterpillar, Komatsu, and Chinese OEMs Tonly and LGMG together account for 88% of the autonomous truck units GlobalData tracks worldwide. China Energy operates the single largest fleet among mining companies, at 509 units, and the Baishihu coal mine holds the largest single-site fleet on record, at 420 trucks.

The country-level growth pattern is uneven enough to be informative on its own. China’s fleet nearly quadrupled between July 2024 and July 2025, while Australia’s — an earlier and more mature adopter, going back to Rio Tinto and Fortescue programs that started in the 2008–2020 period — grew by about 10% over the same span, consistent with a market that is scaling up existing sites rather than converting new ones at the same pace.

Watch on YouTube: Autonomous haulage and copper resource development, on site

Calculator: Autonomous Haulage Conversion Payback

The numbers above answer “how big is this trend”; the tool below answers “does it pencil out for a specific fleet.” Enter your own truck count, costs and expected downtime reduction — nothing is pre-filled with a number from this article.

Copper Workflow Automation

Below the haul road, copper-specific processing has its own automation track. Solvent-extraction/electrowinning (SX-EW) circuits and concentrate smelters have moved toward closed-loop reagent dosing, automated froth monitoring, and digital twins that simulate a plant before an operator changes a setpoint — the same direction of travel as haulage automation, applied to metallurgy instead of transport. Because copper concentrate and cathode grades are proprietary to each operation, there is no equivalent global tracker like GlobalData’s truck count for processing-plant automation; the reliable way to check a specific site’s status is the operator’s own investor disclosures or technical reports, not a market-wide percentage. For current U.S. mine production and price data specifically, the source of record is the USGS National Minerals Information Center’s Mineral Commodity Summaries, published every February — check that year’s copper chapter directly rather than relying on a figure quoted secondhand.

Watch on YouTube: AI drones and hyperspectral mapping over an Arizona porphyry system

Underground Autonomy, Sensor Networks and Digital Twins

Below surface, remotely operated vehicles and robotic drilling and bolting systems extend autonomy into ground that would otherwise require a human in a hazardous heading. Sensor networks feeding real-time digital twins — of ventilation, ground stability, and ore-body geometry — are what make that remote operation auditable rather than a leap of faith: an operations center can see the same conditions a site geologist would see underground, without sending a person down to check.

That same sensor-and-model logic is what satellite-based mineral detection applies before any shaft is sunk. Farmonaut’s satellite-based mineral detection service uses multispectral and hyperspectral imagery with AI to flag alteration zones and structural targets across a region, so ground crews and drill programs go to fewer places before finding the right one.

Watch on YouTube: Satellite-based critical-minerals exploration methods in Canada

Mining Automation and Safety: What MSHA’s Numbers Show

MSHA’s own fiscal-year dashboard — updated twice a year, most recently on February 27, 2026 — records 28 mining fatalities for fiscal year 2025 across an industry covering 327,465 miners: 5 in coal mining, 23 in metal and nonmetal mining. The fatal-injury rate was 0.0101 per 200,000 employee hours worked (coal: 0.0085; metal/nonmetal: 0.0105), and the all-injury rate was 1.77 per 200,000 hours (coal: 2.73; metal/nonmetal: 1.52). For scale, MSHA’s own historical marker is 1978, the first full year under the Federal Mine Safety and Health Act of 1977, when 248 miners died in mining accidents — a decline of more than 88% since, across an industry that automated substantially over the same period.

Automation’s exact contribution to that safety record has no clean number attached to it: MSHA’s public fatality data does not tag incidents by whether the equipment involved was operator-driven or autonomous, so no dataset currently lets you isolate automation’s effect with a figure. What is checkable is the mechanism — an autonomous haul truck removes a person from the cab of the single class of machine that has historically produced the most powered-haulage incidents in surface mining, which is a structural change in exposure, not a statistical claim about this year’s count. Before quoting the 28-fatality figure anywhere, confirm today’s number at MSHA’s Mine Safety and Health at a Glance page, since it is republished on a fixed biannual schedule.

Watch on YouTube: Satellite soil-geochemistry exploration methods in British Columbia

Farmonaut’s Role: Satellite Mineral Intelligence Before a Rig Moves

Autonomous trucks and robotic drills automate what happens once a mine is producing. Farmonaut’s work sits earlier in the timeline: identifying where to explore before a single hole is drilled. Conventional exploration can take years of ground surveys and speculative drilling to rule targets in or out. Farmonaut’s satellite-based approach applies AI to multispectral and hyperspectral imagery to screen whole regions in days, flagging alteration zones and structural targets ahead of ground crews.

Farmonaut states that this approach can lower early-stage exploration costs by 80–85% compared with conventional ground surveys — a claim about its own product, worth verifying against your own program’s baseline rather than treating as an independent industry-wide figure. Explore the satellite-based mineral detection product page for the method, or review the satellite-driven 3D mineral prospectivity mapping approach for subsurface modeling that feeds directly into drill planning.

Ready to scope a program? Get a quote for a mining intelligence report, or contact us to discuss a specific target area.

Watch on YouTube: Satellite-based gold exploration, a regional case study

Watch on YouTube: Satellite-based gold exploration, method overview

Greenhouse Automation News: The Current State

The most recent hard number on U.S. greenhouse-scale production comes from USDA NASS’s 2024 Census of Horticultural Specialties, released February 26, 2026: $18.3 billion in total floriculture, nursery and specialty-crop sales across 23,060 operations. Within that, sales of food crops grown under protected structures — the census category that best matches “greenhouse produce” — reached $1.01 billion, up 44% since 2019 (implying a 2019 baseline of roughly $0.70 billion). California ($3.07B), Florida ($2.15B) and Oregon ($1.29B) led all states, and the top ten states combined for 67% of national sales. Labor made up 36% of total industry expenses, itself up 33% since 2019 — the cost pressure that automation vendors are selling against.

The forward-looking market picture comes from private market-research forecasts rather than a government census, and should be read as a forecast rather than a fact: SNS Insider projects the U.S. controlled-environment agriculture market to grow from $24.72 billion in 2025 to $71.76 billion by 2035, an 11.27% compound annual growth rate, with the global market moving from $82.00 billion to $282.44 billion over the same window (13.20% CAGR). Within that spend, greenhouses hold 54.3% of the facility-type share, hardware holds 45.2% of component spend, and software — the category that includes automation, farm-management and AI-based monitoring platforms — is growing fastest, at a projected 15.19% CAGR.

The past three years have also been a sorting exercise, not a straight growth line. Well-funded ventures including AppHarvest, Plenty, Bowery Farming and Kalera went through bankruptcy, while operators that leaned on automation to control labor cost kept expanding: Gotham Greens opened its twelfth greenhouse — a 210,000-square-foot, highly automated facility in Monroe, Georgia — in October 2023, its first in the Southeast, and Little Leaf Farms raised $300 million in 2022 to fund a 180-acre facility in McAdoo, Pennsylvania. The pattern industry coverage describes for the period ahead is less about building new greenhouse footprint and more about automating, sensor-instrumenting and software-upgrading the footprint that already exists.

On the research side, Wageningen University & Research’s Autonomous Greenhouse Challenge is the closest thing greenhouse automation has to a standardized benchmark. Since its first edition in 2018, international teams have used sensors, cameras and AI to run entire crop cycles — cucumbers, tomatoes, lettuce, and tomatoes again in the 2024–2025 edition — with no human adjusting the climate or irrigation controls, at WUR’s Bleiswijk facility in the Netherlands. Across editions, competing algorithms have matched or beaten commercial cherry-tomato yield benchmarks while using less water and energy. A fifth edition is planned for 2026. This is the durable part of the greenhouse automation story: the method is public, repeatable and gets re-run every year, unlike a single market-size forecast that ages the moment it’s published.

On the policy side, the UAE gives the clearest government-driven example for readers in the Gulf. The National Food Security Strategy 2051 set a target of ranking in the world’s top ten on the Global Food Security Index by 2021 and first by 2051, built on 38 initiatives across 18 priority food categories, with controlled-environment agriculture named as a core pillar for reducing import dependence. ADQ and Netherlands-based Safe Haven Solutions have already put that into a physical building: a 10-hectare automated, climate-controlled greenhouse growing tomatoes and cucumbers at ADQ’s AgTech Park in KEZAD, Abu Dhabi, announced in 2023, with a second phase planned to double the growing area to 20 hectares.

IoT for Plant Automation: The Sensor Layer

Underneath both the WUR competition results and the commercial CEA build-out is the same infrastructure layer: networked sensors measuring temperature, humidity, CO₂, substrate moisture and light, feeding a controller that adjusts vents, irrigation and lighting without a person walking the rows. That is the “IoT for plant automation” market in practice, and it is the reason software is CEA’s fastest-growing spending category at a projected 15.19% CAGR (SNS Insider) — the hardware sensors are increasingly a commodity, and the value has shifted to the models interpreting their data. In open-field agriculture, USDA’s Economic Research Service found a comparable pattern by farm size rather than by facility type: guidance autosteering systems, a proxy for IoT-linked field automation, were used on 52% of midsize farms and 70% of large-scale crop-producing farms in 2023, while yield monitors, yield maps and soil maps reached 68% of large-scale farms, according to the USDA ERS chart published December 10, 2024. Smaller operations, in both greenhouse and field settings, adopt these sensor systems later — largely a function of the fixed cost of instrumenting a space regardless of its size.

Benefits, Risks and What to Check Before You Buy

Set against the numbers above, three benefits are actually documented rather than promised. Labor-cost exposure drops in proportion to headcount removed, which is a real number once you plug your own site’s payroll into the calculator above. Resource use improves under controlled testing: WUR’s Autonomous Greenhouse Challenge editions have repeatedly shown lower water and energy use alongside equal or higher yield, under competition conditions. And equipment utilization improves where autonomous fleets are documented running continuous shift patterns without the fatigue-driven slowdowns that affect human-operated equivalents — though no public dataset in this article quantifies that gain as a percentage you can cite directly.

Three risks deserve equal weight. First, cybersecurity and connectivity dependency: an autonomous fleet or a networked greenhouse controller is a control system, and control systems need the same patching, monitoring and incident-response discipline as any other industrial network — a lapse here stops production, not just a single machine. Second, capital intensity: the calculator above makes conversion capex explicit precisely because it is usually the line item that kills a business case that looked attractive on a labor-savings basis alone. Third, workforce transition: BLS occupational data (SOC 47-5049, Underground Mining Machine Operators) is the place to check current employment and wage figures for roles automation displaces most directly — plan for reskilling into supervisory and maintenance roles rather than assuming headcount simply falls to zero, since remote operations centers still need staff.

Watch on YouTube: Australia’s gold mining automation and sustainability push

FAQ

What does “mining automation” mean in practice, today?

Concretely, it means autonomous or tele-remote haul trucks, drills and loaders — 3,832 such trucks were in operation worldwide in July 2025 per GlobalData’s Mining Intelligence Center, up from 2,080 a year earlier — plus the sensor networks and digital twins that let a remote operations center run them. It does not yet mean a fully unmanned mine; most sites run a hybrid fleet with autonomous and operator-driven equipment side by side.

Is greenhouse automation news actually a live story, or is it settled?

It’s live and unsettled. USDA’s 2024 census shows real growth in protected-structure food-crop sales (+44% since 2019), but the same period saw well-capitalized indoor farming ventures — AppHarvest, Plenty, Bowery Farming, Kalera — go through bankruptcy while automation-focused operators like Gotham Greens and Little Leaf Farms kept expanding. The open question industry coverage keeps returning to is whether automation lowers the labor cost enough to make large indoor-growing footprints profitable, not whether the sensors and AI controllers themselves work — Wageningen University’s competition results say they do.

What is copper workflow automation, specifically?

It refers to automation inside copper processing — solvent-extraction/electrowinning circuits and concentrate smelters — rather than at the mine face. Trends include closed-loop reagent dosing, automated froth monitoring and digital-twin process simulation. There is no global adoption tracker for this specific to copper metallurgy; check individual operators’ technical reports for site-level status.

Does mining automation measurably improve safety?

MSHA’s public fatality data does not separate incidents by whether the equipment involved was autonomous or operator-driven, so there is no dataset that isolates automation’s safety effect as a number. What is verifiable is fiscal-year 2025’s total of 28 U.S. mining fatalities (5 coal, 23 metal/nonmetal) against a 1978 baseline of 248 — and the structural argument that an unmanned truck cab removes a person from the highest-exposure position on the machine.

What role does Farmonaut play in mining automation?

Farmonaut works upstream of on-site automation, applying satellite-based, AI-driven analysis of multispectral and hyperspectral imagery to screen regions for mineral targets before drilling starts. The company states this can lower early-stage exploration costs by 80–85% versus conventional ground surveys, streamlining ground validation ahead of any autonomous haulage or drilling fleet being deployed.

Watch on YouTube: Inside the global gold mining industry

Conclusion

Mining automation and greenhouse automation are separate industries moving on separate timelines, but both are past the pilot stage and both now generate numbers you can check yourself. On the mining side, GlobalData’s fleet counts and MSHA’s fatality data update on fixed schedules — use the links in this article to pull the current figures rather than this article’s snapshot. On the greenhouse side, USDA’s Census of Horticultural Specialties runs again in roughly five years, and Wageningen University’s Autonomous Greenhouse Challenge re-runs annually, so both will keep producing fresh, checkable evidence rather than a single frozen claim.

For the exploration stage that comes before either autonomous haulage or greenhouse sensor networks matter, map your mining site today and see what satellite-based mineral intelligence turns up before committing drilling capital.