SMARTWELL

Three forces converge

Labor scarcity

Aging rural populations and rising wages make manufacturing and agriculture increasingly dependent on robots. Here is how the leaders are commercializing — and where Smartwell runs.al picking, weeding and pollination harder to staff every season — the core driver behind John Deere's autonomy push and China's robotics policy support.

Model maturity

Embodied AI (world models, vision-language-action) crossed lab thresholds in 2024-2026; China's harvest-robot startups are already delivering orders — the industry consensus: the eve of industrialization.

Market size

Global smart-agriculture market: $5.6B (2023) → $13.7B (2029E), CAGR 19.7% (MarketsandMarkets). China's agri remote-sensing market grows at 45.2% CAGR. High-value crops are the beachhead.

Four commercialization archetypes

John Deere · 约翰迪尔

Open-field giant · Retrofit
  • 2nd-gen autonomy kit (CES 2025): 16-camera pods, 360° perception — no lidar
  • Autonomous 9RX tractor + 5ML orchard tractor; CES 2026 extended to X9 combine
  • Retrofit kits via dealer channel (MY2020.5+ 8R / MY2022+ 9R); goal: fully autonomous US corn & soy by 2030

XAG · 极飞科技

Drones-first · IPO filed
  • HKEX H-share filing Sep 2025 — chasing "first ag-drone stock"
  • 2024 revenue ¥1.066B (+128% overseas); drones = 89% of revenue; world #2 behind DJI
  • R100/R200 unmanned ground vehicles + RM80 mower — from human-machine to full robotization

Carbon Robotics

Physical-AI single hit · Specialty
  • LaserWeeder: 5,000+ weeds/min with computer vision + thermal lasers, 10B+ weeds killed; grower weeding cost cut from $1M to ~$0.1–0.2M
  • Crossed $100M revenue; $157M+ funding (BOND, NVIDIA NVentures, Giant Ventures)
  • 2025: AutoTractor — supervised autonomy retrofitted onto existing tractors

China harvest-robot newcomers

Embodied picking · Pre-scale
  • Qiaogeli (Hangzhou): tomato robots with ~4h unattended runs, extending to strawberry/cucumber/pepper
  • Weijing: humanoid picker at ~2s per pick, "one robot ≈ 5–6 workers", order deliveries started
  • Xiong'an dual-arm humanoid tomato picker; university lab fleets — industry at pre-industrialization stage

Software-only SaaS (reference)

Prospera (acquired), Climate FieldView (Bayer), Icanong: strong data & agronomy, but no hardware execution — advice without hands.

General humanoids (Figure · Tesla · Unitree)

Capital and attention concentrate in factories and logistics first; agriculture remains exploratory — structured farming tasks favor specialized forms (dual-arm, wheeled) over general humanoids.

Field × Brain: the 2×2

Open field
CEA / structured
Hardware-first
Hardware-first · Open field
John Deere — autonomy kits, dealer retrofit
XAG — drones + UGVs, IPO filed
Hardware-first · CEA
Harvest newcomers — picking arms, pre-scale
Carbon Robotics — single-task Physical AI
Brain-first
Brain-first · Open field
Yuntong DaaS (ours) — satellite brain at scale
Software SaaS peers — no execution
★ SMARTWELL — Brain-first × CEA × full stack
Dual-engine brain already commercial (CEA + Yuntong cash engines)
Robots as hands inside structured CEA — fastest reliability path
One closed loop: sense → decide → act → learn

Different by construction

The brain is commercial, today

Our dual-engine models run 110+ paying CEA clients and 20+ Yuntong institutional customers — revenue (unaudited), 86%+ gross margin. Robot R&D is funded by real cash engines, not by hope.

Data loop nobody else has

CEA micro-sensing + satellite macro-sensing + robot execution results — three data streams, one brain. Every harvest trains the next. Peers have at most one stream.

A listed company's discipline

TSXV-listed (SMWE) with audited reporting, two CAE academicians advising, CMMI-5 processes and a capital platform for the long R&D cycle — rare among robotics startups.

The right beachhead

High-value CEA crops (strawberry, hemp, tomato) concentrate economics: tight rows, repeatable tasks, premium produce — where a 6× efficiency gain pays back fastest. We already operate there.

Three steps to the fleet

1

Prove in our own greenhouses

Run supervised picking/de-leafing/pollination pilots inside CEA facilities we already serve — 3 regions, 8 structures, 5 orchards.

2

Anchor-customer fleets

Multi-robot scheduling pilots with anchor customers; Nanjing cooperation in planning. RaaS subscriptions begin.

3

Scale as a category

2028 Q4: commercial fleet launch for high-value crops — 6× daily efficiency target, robot + brain + data bundled.


Peer information compiled from public sources (Sep 2026): Deere CES announcements; XAG HKEX filing; Carbon Robotics funding releases; Chinese media coverage of harvest robots. Market sizes: MarketsandMarkets, CYXXW, Taibo reports as cited in the company public disclosures.
Robot program status: R&D stage — demo animations on this site are conceptual, not footage of physical units. Forward-looking statements apply; see footer.