Netafim Deploys AI Irrigation to Halve Farm Water Use, Double Yields

Presentation by Max Moldavsky, Netafim — Ai4

Ai4

Digital farming and predictive crop modeling must replace legacy flood irrigation to satisfy a 70% surge in global food demand as arable land vanishes.

AI-assisted summary of Presentation by Max Moldavsky, Netafim — original source. Check the original for full context.

Source published:

Detailed notes

## Background Published on YouTube by the Ai4 channel on August 28, 2025, this talk features Max Moldavsky, Director of Innovation and Climate Solutions at Orbia Netafim. The presentation description highlights how agtech pioneers are applying artificial intelligence through intelligent irrigation, predictive analytics, and data-driven resource management to shape the future of food, water, and climate resilience.

## TL;DR Max Moldavsky addresses the critical agricultural paradox of needing to produce 60% to 70% more food by 2050 despite depleting arable land, worsening freshwater scarcity, and climate instability. While precision irrigation dramatically reduces water waste, Moldavsky argues that agriculture must now transition to digital farming by optimizing 'every drop of data' through closed-loop systems like Netafim's GrowSphere. By combining six decades of agronomic domain knowledge with real-time sensor telemetry, hyper-localized AI models, and conversational agents, the platform automates crop management from planning to execution. This data-driven transformation democratizes expert agronomy for smallholders worldwide, stabilizes commercial food supply chains, and unlocks fairer agricultural financing.

## Atmosphere & style The presentation carries a professional, urgent, yet pragmatic and tech-forward tone characteristic of an executive agtech keynote. Moldavsky speaks with steady authority and clarity, shifting smoothly from sobering global resource statistics to practical, real-world field applications. The pacing is efficient and optimistic, using concrete operational vignettes to ground complex data architectures in the everyday realities of farmers. It feels like an engaging, high-level briefing on how physical hardware and artificial intelligence converge to address one of humanity's core existential challenges.

## Key takeaways - [00:04](https://www.youtube.com/watch?v=rra9m3VS6e0&t=4s) Global agriculture faces an 'age of less' characterized by the loss of one-third of arable land in 40 years and shrinking freshwater supplies, even as food production must surge 60% to 70% by 2050. - [02:17](https://www.youtube.com/watch?v=rra9m3VS6e0&t=137s) Agriculture accounts for 70% of global freshwater consumption, yet that water is concentrated on only 24% of farmed land—where 75% still relies on ancient, inefficient flood irrigation. - [02:17](https://www.youtube.com/watch?v=rra9m3VS6e0&t=137s) Precision drip irrigation delivers water and nutrients directly to root zones, reducing water consumption by an average of 50% while increasing crop yields by 50% to 100%. - [04:36](https://www.youtube.com/watch?v=rra9m3VS6e0&t=276s) Physical water optimization alone is no longer enough to resolve the food crisis; agriculture must advance into digital farming by optimizing 'every drop of data'. - [04:36](https://www.youtube.com/watch?v=rra9m3VS6e0&t=276s) Operating systems like GrowSphere establish a closed-loop lifecycle that integrates planning, real-time sensing, dynamic AI recommendations, and automated field execution. - [06:02](https://www.youtube.com/watch?v=rra9m3VS6e0&t=362s) The core agricultural AI engine combines decades of empirical agronomic models with plot-level hyper-localization and conversational AI agents that assist or automate field tasks. - [09:17](https://www.youtube.com/watch?v=rra9m3VS6e0&t=557s) Real-time telemetry and remote sensing eliminate manual field scouting friction, dynamically adapt schedules to unexpected rainfall, and benchmark performance against global crop datasets. - [11:46](https://www.youtube.com/watch?v=rra9m3VS6e0&t=706s) AI democratizes professional agronomic expertise for smallholder farmers cultivating single-digit acreages who cannot afford dedicated agronomists. - [11:46](https://www.youtube.com/watch?v=rra9m3VS6e0&t=706s) Data integration stabilizes procurement and processing capacity for food manufacturers while providing end-to-end farm-to-table traceability for consumers. - [14:00](https://www.youtube.com/watch?v=rra9m3VS6e0&t=840s) Objective, AI-derived farm telemetry unlocks fairer crop insurance and financing terms for underserved growers by providing verified risk data to financial institutions. - [14:00](https://www.youtube.com/watch?v=rra9m3VS6e0&t=840s) Hyper-localized predictive modeling overturns rigid, legacy farming calendars by discovering microclimate-optimized planting and harvesting schedules. - [14:00](https://www.youtube.com/watch?v=rra9m3VS6e0&t=840s) The impending data-driven agricultural revolution is poised to exceed the scale of the 20th-century agricultural revolution that expanded global food production tenfold.

## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=rra9m3VS6e0&t=4s) The Paradox of Scarcity and Demand in Global Agriculture

Max Moldavsky, Director of Innovation and Climate at Netafim, introduces his presentation by shifting the focus from standard tech conference fare to how artificial intelligence is actively reshaping agriculture. He establishes the stakes by describing modern humanity as living in an "age of less." First, the world is grappling with shrinking freshwater resources, suffering depletion even in regions traditionally considered climatically secure. Second, arable land suitable for cultivation is vanishing at an alarming rate; over the past 40 years alone, the planet has lost one-third of its arable land. Third, climatic stability has deteriorated, marked by an escalating cadence of extreme weather events—ranging from droughts and catastrophic floods to wildfires and severe storms—that disrupt standard farming cycles.

Against this backdrop of resource depletion lies a striking paradox: while critical inputs are diminishing, the demand for output is surging. Driven by population expansion and shifting socioeconomic dynamics, global food production must increase by 60% to 70% by the year 2050. As Moldavsky emphasizes, conventional farming methods are not only incapable of sustaining this surge under diminishing resources, but they have also become "part of the problem."

- **Arable land depletion:** One-third of global arable land has been lost in the last 40 years. - **Future food demand:** Global agriculture must generate 60% to 70% more food by 2050 to meet population needs. - **Climate volatility:** Intensifying droughts, storms, floods, and wildfires are directly reducing predictable crop yields.

---

[02:17](https://www.youtube.com/watch?v=rra9m3VS6e0&t=137s) Inefficiencies in Global Water Use and the Case for Precision Irrigation

Moldavsky illustrates the structural inefficiency of global food systems by breaking down the mechanics of water allocation. Agriculture accounts for 70% of all global freshwater consumption, leaving residential and industrial usage with the remaining 30%. However, that massive 70% slice of freshwater is applied to only 24% of total cultivated agricultural land. The remaining 76% of farmed land relies entirely on natural rainfall, leaving growers critically vulnerable to weather anomalies: "If it's a good season, that's great. If it's a bad season, they're in trouble. We're in trouble. Everyone is in trouble."

```mermaid flowchart TD accTitle: Global Freshwater and Agricultural Land Allocation subgraph Freshwater["Global Freshwater Distribution"] FW_AG["Agricultural Use<br/>(70% of Freshwater)"] FW_IND["Municipal & Industrial<br/>(30% of Freshwater)"] end

subgraph Land["Cultivated Land Irrigation Methods"] L_RAIN["Rainfed Land<br/>(76% of Cultivated Land)"] L_IRR["Irrigated Land<br/>(24% of Cultivated Land)"] L_FLOOD["Traditional Flood / Inefficient<br/>(75% of Irrigated Land)"] L_PREC["Precision Irrigation<br/>(25% of Irrigated Land)"] end

FW_AG --> L_IRR L_IRR --> L_FLOOD L_IRR --> L_PREC ```

Even more critically, 75% of the irrigated 24% still relies on ancient, inefficient methods—such as flood irrigation—virtually unchanged from practices used thousands of years ago. Precision irrigation addresses this disconnect by delivering water and nutrients directly to plant root zones. By optimizing every drop, precision systems routinely reduce water consumption by an average of 50% while simultaneously increasing crop yields by 50% to 100% on the exact same land footprint. Additional benefits include reduced expenditure on inputs like field labor, pumping energy, and crop protection chemicals. Moldavsky notes that Netafim pioneered this space in 1965 out of sheer necessity in the arid Israeli desert, eventually expanding into a global provider serving millions of growers.

---

[04:36](https://www.youtube.com/watch?v=rra9m3VS6e0&t=276s) From Every Drop of Water to Every Drop of Data: Digital Farming with GrowSphere

Despite decades of hardware innovation in drip and precision irrigation, Moldavsky argues that physical water optimization alone is no longer sufficient to solve the global food crisis. Agriculture must advance into digital farming, expanding the mandate from optimizing "every drop of water" to optimizing "every drop of data." Digital farming equips farmers to synthesize vast streams of on-farm and off-farm data to achieve optimized, autonomous operational decision-making across the entire cultivation lifecycle.

```mermaid flowchart LR accTitle: GrowSphere 360-Degree Closed-Loop Lifecycle subgraph Plan["1. Planning"] P1["Crop & Resource Plans"] end subgraph Monitor["2. Monitoring"] M1["Field & Crop Sensors"] M2["Satellites & Drones"] end subgraph Decide["3. AI Recommendation"] D1["Dynamic Agronomic Models"] D2["Weather Adjustments"] end subgraph Execute["4. Operation"] E1["Automated Controllers"] E2["Valves & Fertigation Units"] end

Plan --> Monitor Monitor --> Decide Decide --> Execute Execute --> Plan ```

To deliver this capability, Netafim developed **GrowSphere**, an end-to-end irrigation and fertigation operating system. GrowSphere manages a 360-degree closed loop: it establishes crop plans, monitors field conditions in real time via heterogeneous sensor networks, generates dynamic operational recommendations for water and fertilizer dosing, and directly executes those commands through automated field controllers.

---

[06:02](https://www.youtube.com/watch?v=rra9m3VS6e0&t=362s) Two-Layer System Architecture and the Core Pillars of Agricultural AI

The GrowSphere system operates across two core layers: 1. **The Hardware Layer:** Encompasses physical hydraulic infrastructure (drip lines, automated valves, filters, dosing units) alongside telemetry sensors measuring soil moisture, plant stress metrics, and localized weather indicators. 2. **The AI-Driven Software Layer:** Ingests real-time field telemetry, remote sensing feeds from satellites and drones, and localized meteorological forecasts into dynamic crop models.

These models generate continuously updated, real-time recommendations. For instance, if an irrigation event was scheduled two days prior but unexpected rain fell overnight, the system dynamically postpones irrigation by three days to prevent saturation and waste.

```mermaid flowchart TD accTitle: GrowSphere Two-Layer System Architecture subgraph HardwareLayer["Layer 1: Field Hardware & Sensing"] H1["Hydraulic Equipment<br/>(Valves, Filters, Pumps)"] H2["In-Field Telemetry<br/>(Soil, Plant & Weather Sensors)"] end

subgraph DataIngestion["Layer 2A: External & Remote Data"] S1["Remote Sensing<br/>(Satellites & Drones)"] S2["External Weather Forecasts"] end

subgraph AIPlatform["Layer 2B: AI Agronomic Intelligence Engine"] M1["AI Crop Models<br/>(60-Year Agronomic Baseline)"] M2["Hyper-Localization Engine<br/>(Plot-Level Soil Calibration)"] M3["AI Conversational Agent<br/>(Autonomous Execution & Q&A)"] end

H2 --> AIPlatform DataIngestion --> AIPlatform AIPlatform --> H1 ```

Moldavsky highlights three key pillars powering their agricultural AI engine: - **AI-Enhanced Crop Models:** Combines Netafim's 60 years of empirical agronomic domain knowledge with predictive machine learning to evaluate complex multi-variable interactions in real time. - **Hyper-Localization:** Overcomes broad agronomic generalizations by tailoring irrigation to specific soil variations and microclimates across distinct plots—even between adjacent fields separated by only a few hundred meters. - **Built-in AI Agents:** Alleviates the cognitive burden on farmers by providing natural language interfaces to explain technical charts, clarify system logic, and autonomously manage field controllers when authorized.

---

[09:17](https://www.youtube.com/watch?v=rra9m3VS6e0&t=557s) GrowSphere in Action: Eliminating Guesswork and Field Fatigue

A short operational video dramatizes the daily friction farmers encounter without automated systems: team members debating conflicting field observations, noting a 3.4°F temperature spike, arguing over whether specific plots (such as plots 12, 15, and 33) are over- or under-irrigated, and dreading a five-hour round trip just to inspect a remote northern section.

The vignette demonstrates how GrowSphere resolves these inefficiencies in real time: - Provides immediate status visibility across all plots from any smartphone, tablet, or PC, removing the need for manual field scouting trips. - Surfaces actionable issue alerts—such as an interrupted irrigation cycle and dry soil readings on Plot 7—and automatically adjusts the following day's schedule to compensate. - Dynamically integrates sudden rainfall events into ongoing irrigation schedules. - Features a **Crop Advisor** module that provides instant operational guidance for newly introduced crop varieties by leveraging performance datasets from identical crops grown worldwide.

Moldavsky follows the clip by recalling feedback from a commercial grower who credited a single automated system alert with identifying a critical plot anomaly that would have otherwise gone completely unnoticed under standard visual inspections.

[11:46](https://www.youtube.com/watch?v=rra9m3VS6e0&t=706s) Multi-Stakeholder Impact: Planet, Smallholders, and the Food Industry

Intelligent irrigation and predictive analytics deliver an all-around positive impact across the entire agricultural ecosystem. For the environment and soil, resource efficiency prevents degradation and actively maintains long-term soil health. For growers, improved yields and profitability provide economic sustainability, which is especially critical given the structure of global farming. The vast majority of the world's farmers are smallholders cultivating plots on a single-digit number of acres. These growers typically lack the financial resources or direct access to hire professional agronomists to guide decisions on water allocation, fertilization, and crop protection. Deploying AI models decouples agronomic expertise from physical, in-person consulting, driving a true "democratization of knowledge and skills" that levels the playing field for growers worldwide.

```mermaid flowchart TD accTitle: Value Chain Impact of Agricultural AI subgraph Core["AI & Data Platform"] A["Hyper-Localized AI Models"] end

subgraph Beneficiaries["Stakeholder Impacts"] B["Planet & Soil<br/>Resource efficiency & soil health"] C["Smallholders<br/>Democratized agronomy & higher yields"] D["Food Industry<br/>Predictable supply & capacity planning"] E["Consumers<br/>Farm-to-table traceability"] end

A --> B A --> C A --> D A --> E ```

Beyond individual growers, commercial food processors and agribusinesses benefit from increased predictability across their procurement networks. A core operational KPI for food manufacturers is keeping processing facilities running at full capacity with minimal overhead; real-time visibility into contracted farms stabilizes supply chains and mitigates disruptions. Furthermore, these integrated data streams enable complete farm-to-table traceability, giving end consumers transparent visibility into how their food was grown and handled throughout its lifecycle.

- **Agronomic Access**: Bridges the knowledge gap for smallholder farmers managing single-digit acreage without access to dedicated agronomists. - **Supply Chain Optimization**: Enables food companies to optimize processing plant utilization and procurement planning. - **Consumer Traceability**: Facilitates end-to-end transparency across the food supply chain from harvest to consumption.

---

[14:00](https://www.youtube.com/watch?v=rra9m3VS6e0&t=840s) The Broader Frontier: Smarter Finance, Rewriting Farming Dogmas, and the Next Revolution

While precision irrigation serves as a prime proof point, the potential applications of AI across agriculture extend into broader structural domains. Financial and insurance institutions have historically struggled to underwrite small or non-standard agricultural operations due to lack of verifiable risk data. AI-driven predictive modeling provides credible, objective farm-level data that enables financial institutions to offer improved financing terms and fairer insurance premiums to growers who were previously uninsurable or underserved.

AI also has the potential to challenge and overturn deeply entrenched agricultural dogmas. Traditional farming calendars often rely on broad regional heuristics—such as the standard rule of thumb to plant corn in March or April and harvest in September or October. Hyper-localized AI models can evaluate microclimates, soil dynamics, and shifting weather patterns to discover entirely new cultivation cycles that maximize productivity beyond legacy practices.

This impending transformation is poised to surpass the scale of the early 20th-century agricultural revolution, which expanded global food production tenfold. By harnessing data-driven decision-making, the next agricultural revolution will center on empowering millions of farmers to optimize "every drop of data" alongside every drop of water.

- **Financing and Risk Underwriting**: High-fidelity AI farm models allow insurers and lenders to offer lower premiums and accessible capital. - **Challenging Conventional Calendars**: Hyper-local modeling can replace rigid seasonal conventions (e.g., standard spring-planting and autumn-harvest corn windows) with optimized, data-backed schedules. - **Historical Scale**: Anticipated to exceed the tenfold production gains achieved during the 20th-century agricultural revolution.

## Notable quotes > [00:04](https://www.youtube.com/watch?v=rra9m3VS6e0&t=4s) age of less — Max Moldavsky

> [00:04](https://www.youtube.com/watch?v=rra9m3VS6e0&t=4s) part of the problem — Max Moldavsky

> [02:17](https://www.youtube.com/watch?v=rra9m3VS6e0&t=137s) If it's a good season, that's great. If it's a bad season, they're in trouble. We're in trouble. Everyone is in trouble. — Max Moldavsky

> [04:36](https://www.youtube.com/watch?v=rra9m3VS6e0&t=276s) every drop of water — Max Moldavsky

> [04:36](https://www.youtube.com/watch?v=rra9m3VS6e0&t=276s) every drop of data — Max Moldavsky

> [04:36](https://www.youtube.com/watch?v=rra9m3VS6e0&t=276s) GrowSphere — Max Moldavsky

> [09:17](https://www.youtube.com/watch?v=rra9m3VS6e0&t=557s) Crop Advisor — Max Moldavsky

> [11:46](https://www.youtube.com/watch?v=rra9m3VS6e0&t=706s) democratization of knowledge and skills — Max Moldavsky

## Mentioned Max Moldavsky, Netafim, Orbia Netafim, GrowSphere, Crop Advisor, Ai4

## Video description Max Moldavsky, Director of Innovation and Climate Solutions, Orbia Netafim

Abstracts

AI is transforming agriculture, enabling precision, prediction, and resilience at scale. In this talk, we’ll explore how agtech pioneers like Orbia Netafim are applying AI through intelligent irrigation, predictive analytics, and data-driven resource management to shape the future of food, water, and climate.

[Source](https://www.youtube.com/watch?v=rra9m3VS6e0)

中文简报

Netafim部署AI灌溉：农田用水减半、产量翻倍 在耕地锐减的背景下，数字农业与预测性作物建模必须取代传统漫灌，以应对全球激增70%的粮食需求。 Max Moldavsky 阐述了当前严峻的农业悖论：尽管面临耕地缩减、淡水匮乏加剧以及气候不稳定的挑战，到2050年全球粮食产量仍需提升60%至70%。虽然精准灌溉显著降低了水资源浪费，但 Moldavsky 认为农业如今必须通过类似 Netafim 的 GrowSphere 等闭环系统来利用好“每一滴数据”，向数字农业转型。该平台将六十年的农学专业知识与实时传感器遥测技术、超本地化AI模型及对话式智能体相结合，实现了作物从规划到执行的全流程自动化管理。这一数据驱动的变革不仅使全球小农户能够获得专家级农学指导，还稳定了商业粮食供应链，并开启了更加公平的农业融资渠道。 全球农业正步入“匮乏时代”，过去40年间已丧失三分之一的可耕地，而到2050年全球粮食产量需增加60%至70%；占全球淡水消耗70%的农业中仍有75%依赖漫灌，而精准滴灌技术可在减少50%耗水的同时使产量提升50%至100%。 (0:04) 单纯优化物理用水已无法应对粮食危机，农业正在向“优化每一滴数据”的数字农业转型；Netafim 开发的 GrowSphere 系统构建了涵盖规划、实时传感、动态 AI 建议与自动化执行的端到端闭环。 (4:36) 农业 AI 引擎融合了60年的实证农艺模型、地块级超本地化校准以及 AI 智能体交互，可结合突发降雨等实时遥测与卫星数据动态调整灌溉排期，彻底消除人工巡田的滞后与误差。 (6:02) AI 的应用打破了专业咨询壁垒，让占据全球绝大多数、无力聘请独立农艺师的小农户也能获得精准指导，同时助力食品供应链实现全程可追溯，并为农业金融与保险提供可靠的风险量化数据。 (11:46)

Reading summary

Max Moldavsky of Netafim argues that overcoming the global food crisis amid worsening resource scarcity requires transitioning from physical water efficiency alone to closed-loop digital farming governed by hyper-localized artificial intelligence. Resource Depletion and Irrigation Inefficiencies (0:04) Moldavsky frames modern agriculture as operating in an "age of less," defined by diminishing freshwater resources, an increasing cadence of extreme weather events, and the loss of one-third of global arable land over the past 40 years. In contrast, population growth requires agricultural output to expand by 60% to 70% by 2050. Conventional practices are structurally inadequate to handle this demand, particularly given historical imbalances in water allocation. Globally, agriculture accounts for 70% of freshwater consumption, but that share is applied to only 24% of cultivated land, leaving the remaining 76% vulnerable to erratic rainfall. Furthermore, 75% of irrigated land still relies on ancient, inefficient methods like flood irrigation. According to Moldavsky, precision irrigation resolves this gap by targeting plant root zones directly, which reduces water consumption by an average of 50% while increasing crop yields by 50% to 100% on the same land footprint alongside lowering labor, energy, and chemical costs. Closed-Loop Digital Farming and AI Models (4:36) While precision irrigation hardware provides physical efficiency, Moldavsky asserts that food security requires optimizing "every drop of data" through digital farming. Netafim's GrowSphere operating system coordinates a 360-degree closed loop: establishing crop plans, monitoring field conditions, delivering automated dosing recommendations for water and fertilizer, and executing commands directly through field controllers. The system operates via two integrated layers. The hardware layer gathers field telemetry—including soil moisture, plant stress metrics, and localized weather data—alongside automated hydraulic components. The software layer ingests this telemetry with satellite imagery, drone feeds, and weather forecasts into dynamic crop models built on 60 years of agronomic data. By calibrating to micro-variations between adjacent plots and deploying conversational AI agents to clarify decisions or manage controllers, the system dynamically reschedules irrigation events when unexpected rainfall occurs. Operational Efficiency and Democratizing Agronomic Expertise (11:46) In daily management, integrated digital platforms eliminate manual field scouting and disputes over plot anomalies, such as sudden temperature spikes or interrupted irrigation cycles. In addition, automated crop advisor modules supply operational guidelines for unfamiliar crop varieties by referencing international performance datasets from identical crops. Crucially, Moldavsky highlights that software-driven agronomy democratizes specialized expertise for smallholder farmers cultivating single-digit acreages. Because smallholders generally lack the financial capital to retain professional agronomists, automated AI models provide equitable access to technical guidance. Beyond farm borders, real-time field data offers commercial food processors predictable procurement streams to keep processing facilities running at full capacity, while establishing farm-to-table traceability for consumers. Transforming Agricultural Underwriting and Cultivation Dogmas (14:00) Beyond irrigation, predictive farm modeling provides verified, objective operational data that allows insurers and lenders to underwrite small or non-standard farms that were previously excluded or underserved, opening pathways to fairer premiums and capital access. Furthermore, hyper-localized AI models challenge rigid agricultural conventions, such as generalized regional calendars prescribing spring planting and autumn harvests for crops like corn. By analyzing plot-level microclimates and shifting meteorological conditions, algorithms can surface optimized, non-traditional cultivation windows. Moldavsky forecasts that the scale of this data-driven agricultural transformation will surpass the early 20th-century agricultural revolution, which increased global food production tenfold. Moldavsky concludes that pairing precision irrigation infrastructure with hyper-localized AI models provides a viable framework to expand global agricultural yields under severe resource constraints. Realizing this projected revolution, however, hinges on broad adoption among resource-constrained smallholders and the willingness of agricultural and financial institutions to replace entrenched heuristics with automated predictive systems.

精炼阅读摘要

Max Moldavsky 指出，面对全球耕地急剧流失与未来粮食需求激增的双重危机，传统粗放农业已难以为继；农业生产必须从仅优化“每一滴水”转向优化“每一滴数据”，依托两层式智能操作系统实现精准闭环，从而推动全产业链变革并重塑百年耕作范式。 全球水土供需悖论与灌溉结构失衡 (0:04) Moldavsky 表示，现代农业正身处严峻的“紧缩时代”。在过去40年中，全球已有三分之一的可用耕地遭到侵蚀与破坏，同时频发的干旱、严重洪涝、野火和风暴正持续削弱气候稳定性。与资源锐减形成鲜明反差的是，受人口增长驱动，全球粮食产出至2050年必须提高60%至70%。Moldavsky 强调，依赖传统农业模式不仅无法弥补这一缺口，其低效的耕作机制本身也在加速资源耗竭。 水资源利用的结构性矛盾尤为突出：农业消耗了全球70%的淡水储量，但该份额仅覆盖了全球24%的耕地，其余76%的土地完全依赖天然降雨，极易受极端天气冲击。而在已实施灌溉的24%耕地中，仍有75%沿用着数千年前的大水漫灌等粗放方式。精准滴灌技术将水分与养分直接输送至植物根部，据统计可在同等土地面积上降低50%的平均耗水量，同时使作物产量提升50%至100%，并同步减少人工、能耗及农药化肥支出。 从节水到数据闭环：GrowSphere架构 (4:36) Moldavsky 认为，仅凭硬件层面的水肥节约已无法彻底化解农业危机，现代农业必须步入数字农业阶段，把目标由利用好“每一滴水”升维为利用好“每一滴数据”。为此，Netafim 开发了名为 GrowSphere 的灌溉与施肥操作系统，构建起涵盖作物规划、实时监测、AI推荐决策与自动化执行的360度闭环管理链路，帮助种植者依靠海量数据实现全流程的自主操作。 该系统由两层架构驱动：底层为集成滴灌管网、自动化阀门、施肥机以及测量土壤水分、植株胁迫与田间气象的硬件遥测层；上层为汇聚传感器、卫星无人机遥感与外部天气预报的 AI 软件层。该算法依托三大技术支柱：一是融合 Netafim 六十年农艺实证积累与机器学习的作物模型；二是实现地块级的超本地化校准，即使相距仅数百米且土质不同的相邻田块也能单独运算；三是内置对话式 AI 代理，能以自然语言解释图表逻辑并在授权后自主调度田间阀门控制器。 消除田间误判与复杂场景下的自主调度 (9:17) 在缺乏数字化工具的传统农田管理中，管理者往往深陷于主观推测与沟通内耗。例如，巡查人员常因气温上升3.4华氏度等局部波动对不同地块是否缺水产生分歧，有时为了查看偏远地块甚至需要长达5小时的往返车程。此类依赖肉眼观察的经验主义操作，极易漏判早期的土壤干旱或设备故障。 GrowSphere 通过移动端与电脑端提供全地块的统一可视化看板，免去了耗时的人工巡检。系统能主动捕捉异常事件——如识别出7号地块灌溉中断导致土壤干旱，并自动追加次日的灌溉任务进行补偿；若遇突发降雨，系统可自动将原定灌溉顺延数日，防止积水沤根与水资源浪费。此外，系统的作物顾问（Crop Advisor）模块还能调取全球同类作物的运行数据库，为农场引种的新品类提供即时操作指南。 农艺平权、产业链协同与范式重塑 (11:46) 智能灌溉与预测分析的效益正外溢至整条农业链条。全球绝大多数种植者是耕作面积仅有几个英亩的小农户，通常无力承担专属农艺师的咨询费用。AI 作物模型的普及使高阶农艺知识与实体专家脱钩，推动了专业经验的“平民化”，让小农户得以获得科学配方并改善土壤健康。对于食品加工企业而言，透明的农田数据能稳定原料供应，保证加工厂的产能利用率，并为消费者建立从农田到餐桌的全流程溯源。 在宏观产业层面，Moldavsky 预测 AI 数据模型将改变农业金融与保险体系，通过提供客观透明的风险评估依据，为过去缺乏信用数据的边缘农户争取到更合理的信贷与保费。同时，超本地模型能够瓦解“玉米必须在春季播种、秋季收获”等僵化的传统经验日历，根据气象与土壤动态规划更优的轮作周期。Moldavsky 断言，这场以数据驱动决策为核心的新变革，其增产潜力和深远影响有望超越20世纪初将粮食产量提升十倍的传统农业革命。 Moldavsky 描绘了农业通过数据与 AI 实现全自动化决策的愿景，但这一转变高度依赖田间硬件感知网络与跨区域算法的普适性验证；对于技术基础薄弱且资金受限的小农群体，前期数字基础设施的改造成本与落地难度仍是未完全解答的现实挑战。

Detailed notes — zh

## Background 该视频由 Ai4 频道于 2025 年 8 月 28 日发布在 YouTube 平台。内容记录了 Orbia Netafim 创新与气候解决方案总监 Max Moldavsky 的主题演讲，介绍农业科技领军企业如何通过智能灌溉、预测分析和数据驱动的资源管理等 AI 技术，重塑全球粮食、水资源与气候的未来格局。 ## TL;DR 演讲直面全球淡水萎缩、耕地锐减与人口暴增带来的粮食危机，指出传统大水漫灌模式已不可持续。Netafim 分享了从 60 年前发明滴灌技术迈向数字农业的历程，重点剖析了 GrowSphere 操作系统如何依托 AI 增强农艺模型、超本地化自适应与内置 AI 智能体实现灌溉与水肥一体化闭环控制。该技术不仅填补了小农户缺乏专业农艺指导的鸿沟，还为食品加工业提供了高透明度与供应链确定性。最终，AI 预测能力正在向农业金融、农险风控及种植范式重塑等全产业链延展，驱动农业从“优化每一滴水”跨越到“优化每一滴数据”。 ## Atmosphere & style 演讲整体呈现出兼具严峻危机感与务实科技愿景的专业基调，开篇以一系列悬殊的资源消耗数据迅速抓住观众注意力。主讲人语调沉稳笃定，通过田间混乱管理的生动对比与实机自愈演示，将复杂的农艺算法转化为清晰易懂的工程实践。整场分享节奏流畅紧凑，既有对全球小农户生存处境的人文关怀，又充满引领下一次农业革命的宏大技术自信。 ## Key takeaways - [00:04](https://www.youtube.com/watch?v=rra9m3VS6e0&t=4s) 全球农业正面临水资源减少、可耕地流失与极端天气频发的“匮乏时代”，然而到2050年全球粮食需求将大幅增长60%至70%。 - [00:04](https://www.youtube.com/watch?v=rra9m3VS6e0&t=4s) 传统农业水资源利用极其低效：全球70%的淡水消耗仅灌溉了24%的耕地，且已灌溉耕地中有75%仍沿用传统的大水漫灌方式。 - [03:17](https://www.youtube.com/watch?v=rra9m3VS6e0&t=197s) 精准灌溉不仅能实现50%以上的节水并使作物产量翻倍，更开启了从物理层面的“优化每一滴水”向数字层面的“优化每一滴数据”的跃升。 - [04:36](https://www.youtube.com/watch?v=rra9m3VS6e0&t=276s) GrowSphere 农业操作系统打通田间传感器、卫星遥感与气象预报，构建起涵盖数据分析、模型推演与硬件自动执行的农田管理闭环。 - [07:17](https://www.youtube.com/watch?v=rra9m3VS6e0&t=437s) AI 增强的农艺模型与超本地化（Hyper-Localization）算法相结合，能够针对每一块微地块的独特土壤与微气候定制个性化灌溉方案。 - [07:17](https://www.youtube.com/watch?v=rra9m3VS6e0&t=437s) 内置 AI 农艺智能体通过自然语言交互降低了数字化工具的操作门槛，让农户能轻松理解复杂数据并授权系统自动调度设备。 - [09:11](https://www.youtube.com/watch?v=rra9m3VS6e0&t=551s) 智能监测与自适应决策系统消除了传统农场的混乱管理，实现远程全局监控、灌溉计划智能自愈及突发灾害天气的自动响应。 - [11:46](https://www.youtube.com/watch?v=rra9m3VS6e0&t=706s) AI 推动了农艺专业知识与决策能力的民主化普及，让无力聘请专业专家的广大微型小农户也能享受到顶级农艺算法指导。 - [11:46](https://www.youtube.com/watch?v=rra9m3VS6e0&t=706s) 端到端田间数据为下游食品加工业提供了高可预测的供应链保障，并让消费者得以实现“从农田到餐桌”的全流程透明追溯。 - [14:01](https://www.youtube.com/watch?v=rra9m3VS6e0&t=841s) 高精度的 AI 预测模型为农业金融与保险机构建立了可靠的风控画像，有助于改善小农户的信贷准入条件并降低保险费率。 - [14:01](https://www.youtube.com/watch?v=rra9m3VS6e0&t=841s) 超本地化数据分析正在颠覆千百年来固化的传统农耕经验与种植日历，开启一场深度超越20世纪“绿色革命”的数据驱动农业变革。 ## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=rra9m3VS6e0&t=4s) 农业面临的严峻挑战：资源匮乏与人口增长的双重夹击 Max Moldavsky（Orbia Netafim 创新与气候解决方案总监）在演讲开场时向观众致意，并指出接下来的分享将与过去三天大会上的常规议题有所不同，重点聚焦于人工智能如何引领农业领域的下一次革命。作为全球最大的灌溉公司，Netafim 的故事必须从当前人类面临的核心危机——“匮乏时代（Age of Less）”讲起。这个时代包含三个层面的持续缩减： - **更少的水资源**：全球可用淡水逐年萎缩，水危机甚至在许多以往看似水资源充足的地区蔓延； - **更少的可耕地**：在过去 40 年中，全球已经失去了三分之一适合农业耕作的肥沃土地； - **更低的稳定性**：极端天气事件（干旱、洪涝、野火、风暴）愈发频繁，严重摧毁着土地的生产力。 令人深感矛盾的是，在资源全面走向匮乏的同时，人类需要养活的人口却在持续激增。随着全球人口增长与社会结构变迁，预计到 2050 年，全球粮食需求将增加 60% 至 70%。若继续沿用现有的农业种植模式，不仅无法达成目标，甚至本身就是环境恶化的推手。 数据清晰地揭示了传统农业在水资源与土地利用上的极端低效： - 全球淡水消耗中，高达 **70%** 用于农业灌溉，家庭生活和工业制造仅占 30%； - 然而，这 70% 的农业用水却仅覆盖了全球 **24%** 的耕地； - 其余 **76%** 的农田依然完全“靠天吃饭”，极易受到干旱或异常气候的致命打击； - 更严峻的是，在已经实现灌溉的 24% 土地中，竟有 **75%** 依然沿用数百甚至数千年前祖先使用的传统大水漫灌方式，效率极低。 --- ### [03:17](https://www.youtube.com/watch?v=rra9m3VS6e0&t=197s) 精准灌溉：从优化“每一滴水”起步 面对水资源的巨大浪费与产出瓶颈，“精准灌溉（Precision Irrigation）”应运而生。精准灌溉的核心在于精细化管理“每一滴水”，在最大限度减少资源消耗的同时，实现单位水资源与单位土地面积产出的最大化。通过转向先进的精准灌溉系统，农田平均可节约 50% 以上的水资源，在某些作物上节水比例甚至更高；与此同时，相同地块上的作物产量能够提升 50% 乃至 100%。此外，精准灌溉还大幅削减了劳动力、能源及植保农药等农资投入成本，显著改善了农业的生态可持续性。 Netafim 投身这一领域已有 60 年之久。1965 年，Netafim 诞生于以色列干旱严酷的内盖夫沙漠，正是迫于极度缺水无法种植庄稼的生存危机，才开创了现代滴灌技术。如今，Netafim 已成长为服务全球数百万农户的跨国企业。然而，Max 强调，仅仅优化物理层面的水资源利用仍远远不够，农业必须迈出关键的下一步——拥抱“数字农业革命（Digital Farming Revolution）”，实现从“优化每一滴水”向“优化每一滴数据”的跨越。 --- ### [04:36](https://www.youtube.com/watch?v=rra9m3VS6e0&t=276s) 数字农业大脑：GrowSphere 操作系统架构 数字农业赋予农户整合农场内部与外部所有数据的能力，从而优化资源调配，实现从经验决策向最佳自主决策（Autonomous Decision-Making）的升级。Netafim 为此打造了名为 **GrowSphere** 的灌溉与水肥一体化操作系统。GrowSphere 实现了贯穿农田全生命周期的 360 度闭环管理：从初期规划、多源传感器实时监测、基于数据的实时灌溉与施肥建议，再到直接对接控制硬件自动执行操作指令。 ```mermaid flowchart TD accTitle: GrowSphere 农业操作系统闭环架构 subgraph S1["感知与数据输入层"] A["农田硬件与传感器<br/>(土壤/植物/微气象)"] B["外部数据源<br/>(卫星遥感/无人机/天气预报)"] end subgraph S2["AI 分析与决策层"] C["AI 农艺作物模型<br/>(Crop Models)"] D["超本地化自适应引擎<br/>(Hyper-Localization)"] E["内置 AI 农艺智能体<br/>(AI Agent)"] end subgraph S3["闭环控制与执行层"] F["动态推荐生成<br/>(实时按需调整)"] G["灌溉系统自动化执行<br/>(控制器/阀门/水泵)"] end A --> C B --> C C --> D D --> E E --> F F --> G G -. 状态反馈.-> A ``` GrowSphere 系统整体由两大核心层构成： 1. **硬件层**：包含田间实际运行的灌溉水力管网、过滤器、控制阀门等物理设备，以及部署于土壤、作物本体和周边微气候环境中的各类高精度传感器； 2. **AI 驱动的软件层**：汇集现场传感器数据，深度融合卫星遥感图像、无人机巡检和气象预报等外部资源，输入至核心作物模型（Crop Models）中进行动态运算，生成与现实环境无缝同步的动态决策。例如，若系统两天前计划今天浇水，但昨夜突降甘霖，系统便会自动推迟灌溉计划至 3 天后。 --- ### [07:17](https://www.youtube.com/watch?v=rra9m3VS6e0&t=437s) 赋能农业的三大核心 AI 要素 Max 深入拆解了系统背后的三大核心 AI 支柱，说明技术如何解决传统农业决策中的固有痛点： 1. **AI 增强的农艺模型（AI-Enhanced Agronomical Models）**：Netafim 积累了长达 60 年的宝贵农艺知识与实验数据，但在复杂的自然变量面前依然需要技术升维。AI 能够对海量数据进行模式挖掘与实时推演，大幅强化并拓宽传统农艺知识的应用边界； 2. **超本地化适配（Hyper-Localization）**：传统的农艺建议往往偏向宏观与泛化，但实际农田中“每块土地、每捧土壤都截然不同”。即使是仅相隔几百米的相邻地块，土壤质地与持水特性也可能天差地别。AI 具备针对每一块特定微地块持续学习的能力，量身定制最精准的灌溉配方； 3. **内置 AI 农艺智能体（Built-in AI Agent）**：务农是一项高度复杂的系统性工作，农户必须掌握何时播种、选种、灌溉、施药等海量跨学科知识。AI 智能体通过自然语言交互，帮助农户轻松解读复杂的图表与数据曲线，用最简明直观的方式呈现行动建议，并能在获得授权后代农户自动调度和控制田间设备。 --- ### [09:11](https://www.youtube.com/watch?v=rra9m3VS6e0&t=551s) 田间实测与农户体验：告别混乱决策 一段现场演示视频生动还原了传统农场管理的混乱与焦虑：由于气温突增 3.4 度，21 号地块作物干渴枯萎，而 12、15 和 33 号地块的土壤湿度却严重超标；农场工人们激烈争论当下到底处于“过度灌溉”还是“灌溉不足”，甚至不得不驱车 5 个小时前往偏远的北区地块巡检。 GrowSphere 的介入彻底颠覆了这种低效作业模式： - **全局远程掌控**：管理者无需顶着烈日驱车前往北区，通过电脑、平板或手机即可随时随地查看所有地块的实时灌溉与水肥状态； - **智能告警与排程自愈**：当 7 号地块由于故障未完成预定灌溉、土壤传感器发出干旱告警时，GrowSphere 会立即重新计算并自动修正次日的灌溉计划； - **突发天气与新品种应对**：系统自动将突降暴雨等气象突变纳入灌溉模型，且其内置的“作物顾问（Crop Advisor）”能调取全球类似作物的海量生长数据，帮助农户轻松尝试从未种植过的新品种。 Max 在视频结束后分享了一位真实农户的反馈：该农户曾感慨，GrowSphere 系统发出的一次关键预警，让他及时发现了某块农田潜藏的严重异常，而依靠传统肉眼巡检，“靠自己是绝无可能提前发现这一隐患的”。 ### [11:46](https://www.youtube.com/watch?v=rra9m3VS6e0&t=706s) 全方位共赢：AI 驱动农业知识普惠与端到端供应链赋能 通过实时监测与智能决策系统，农户能够及时识别潜在危机并挽救整片地块，避免颗粒无收的经济灾难。Orbia Netafim 致力于推动的不仅是单点技术的革新，更是构建一个让生态系统各方均能获益的“全方位共赢”（all around positive impact）格局。在生态层面，精准、高效的资源管理直接改善并维护了土壤健康；在农户层面，精准农业显著提升了作物产量与种植收益，使粮食生产在经济与生态双重维度上实现可持续性。 更为关键的是对全球庞大小农户群体的赋能。全球绝大多数农民属于耕作面积仅有几英亩（single-digit acres）的小农户，他们既没有充足的资金，也缺乏渠道去聘请专业农艺师来指导灌溉、施肥、植保等成百上千项繁杂的田间管理决策。AI 模型的介入摆脱了对高成本线下专家巡检的依赖，推动了“知识与技能的民主化普及”（democratization of knowledge and skills），让普通小农户也能拥有顶级专家的决策能力，在同等基准线上获取最佳收成。 ```mermaid flowchart TD accTitle: AI-Driven Multi-Stakeholder Value Flow in Agriculture subgraph Production["田间生产层"] A["环境与土壤<br/>减少资源浪费 / 改善土壤健康"] B["小农户赋能<br/>打破专家门槛 / 知识普惠"] C["种植效益<br/>保障作物产量 / 提高经济收益"] end subgraph Industry["产业与消费层"] D["食品加工业<br/>供应链高可预测性 / 产能利用率优化"] E["终端消费者<br/>从农田到餐桌全程可追溯"] end Production --> Industry ``` 在产业链下游，食品加工业与终端消费者同样迎来深远变革。食品加工巨头最为核心的关键绩效指标之一是以最低成本确保加工厂满负荷运转（"keeping the plant full with minimum costs"），AI 赋予了加工企业对上游签约农场全流程透明化的监控能力，极大增强了原料供应链的韧性与可预测性。对于终端消费者而言，全链条数据实现了“从农田到餐桌”（from farm to table）的全程透明可追溯，让大众清晰了解入口食物在整个生命周期中的生长与管理过程。 - **小农户赋能**：打破传统农艺咨询的高昂资金门槛，让微型地块农户平等获取灌溉、施肥与植保的最优算法指导。 - **供应链可预测性**：帮助食品加工企业实时掌控原料产出进度，降低因原料短缺或过剩导致的工厂停工与损耗。 - **端到端可追溯**：打通田间数据链条，为消费端提供完全透明的食品安全与产地履历保障。 --- ### [14:01](https://www.youtube.com/watch?v=rra9m3VS6e0&t=841s) 超越精准灌溉：重塑农业金融、种植范式与下一场农业革命 精准灌溉仅仅是人工智能赋能农业的一个切入点，AI 在更广泛的农业生态中蕴藏着巨大的应用潜力。以至关重要的农业金融与保险机构为例，以往由于缺乏可靠的田间风险与产量预测模型，农户往往难以获得公允的贷款审批或可负担的农业保险。随着高可信度 AI 预测模型的建立，金融与保险机构能够更精准地评估农场风险画像，从而为广大农户提供更优质的融资渠道与更具竞争力的保险费率。 更深层次的变革在于颠覆人类固有的传统农耕认知。千百年来，农业生产高度依赖代代相传的经验法则，例如人们普遍认为玉米必须在三四月份播种、在九十月份收获。然而借助超高精度的超本地化（hyper-localized）AI 建模与多维气象土壤数据分析，传统种植日历可能被彻底重写，系统能够针对特定微气候环境测算出超越经验直觉的全新轮作与播种窗口，最大化释放土地潜能。 20 世纪初的农业革命（绿色革命）通过机械化与品种改良将粮食产量提升了整整十倍，而当前正在发生的 AI 农业革命其广度与深度将超越以往任何一次历史飞跃。演讲最后，演讲者指出 Orbia Netafim 正在引领这场由人工智能驱动的未来粮食革命，致力于帮助全球数百万农户不仅精准利用好每一滴水，更学会“优化他们所拥有的每一滴数据”（optimize every drop of data they have）。 - **金融与保险创新**：基于可信 AI 模型为农户定制精准风控方案，解决以往因数据缺失导致的信贷难、保费高痛点。 - **种植范式重塑**：打破“春播秋收”等传统经验教条，利用超本地化动态模型定制最优种植与采收窗口期。 - **时代级产业跃升**：超越 20 世纪实现产量增长 10 倍的绿色革命，以“数据驱动”开启全球粮食生产与资源优化的全新篇章。 ## Notable quotes > [00:04](https://www.youtube.com/watch?v=rra9m3VS6e0&t=4s) 匮乏时代（Age of Less） — Max Moldavsky > [00:04](https://www.youtube.com/watch?v=rra9m3VS6e0&t=4s) 更少的水资源：全球可用淡水逐年萎缩，水危机甚至在许多以往看似水资源充足的地区蔓延 — Max Moldavsky > [00:04](https://www.youtube.com/watch?v=rra9m3VS6e0&t=4s) 更少的可耕地：在过去 40 年中，全球已经失去了三分之一适合农业耕作的肥沃土地 — Max Moldavsky > [03:17](https://www.youtube.com/watch?v=rra9m3VS6e0&t=197s) 实现从“优化每一滴水”向“优化每一滴数据”的跨越 — Max Moldavsky > [07:17](https://www.youtube.com/watch?v=rra9m3VS6e0&t=437s) 每块土地、每捧土壤都截然不同 — Max Moldavsky > [09:11](https://www.youtube.com/watch?v=rra9m3VS6e0&t=551s) 靠自己是绝无可能提前发现这一隐患的 — Max Moldavsky > [11:46](https://www.youtube.com/watch?v=rra9m3VS6e0&t=706s) 全方位共赢（all around positive impact） — Max Moldavsky > [11:46](https://www.youtube.com/watch?v=rra9m3VS6e0&t=706s) 知识与技能的民主化普及（democratization of knowledge and skills） — Max Moldavsky > [11:46](https://www.youtube.com/watch?v=rra9m3VS6e0&t=706s) keeping the plant full with minimum costs — Max Moldavsky > [11:46](https://www.youtube.com/watch?v=rra9m3VS6e0&t=706s) 从农田到餐桌（from farm to table） — Max Moldavsky > [14:01](https://www.youtube.com/watch?v=rra9m3VS6e0&t=841s) optimize every drop of data they have — Max Moldavsky ## Mentioned Max Moldavsky, Orbia Netafim, Netafim, GrowSphere, Crop Models, AI Agent, Crop Advisor, 绿色革命 (Green Revolution), Ai4 ## Video description Max Moldavsky, Director of Innovation and Climate Solutions, Orbia Netafim Abstracts AI is transforming agriculture, enabling precision, prediction, and resilience at scale. In this talk, we’ll explore how agtech pioneers like Orbia Netafim are applying AI through intelligent irrigation, predictive analytics, and data-driven resource management to shape the future of food, water, and climate. [Source](https://www.youtube.com/watch?v=rra9m3VS6e0)

Source quotes

[0:04] Max Moldavsky: 匮乏时代（Age of Less） [0:04] Max Moldavsky: 更少的水资源：全球可用淡水逐年萎缩，水危机甚至在许多以往看似水资源充足的地区蔓延 [0:04] Max Moldavsky: 更少的可耕地：在过去 40 年中，全球已经失去了三分之一适合农业耕作的肥沃土地 [3:17] Max Moldavsky: 实现从“优化每一滴水”向“优化每一滴数据”的跨越 [7:17] Max Moldavsky: 每块土地、每捧土壤都截然不同 [9:11] Max Moldavsky: 靠自己是绝无可能提前发现这一隐患的 [11:46] Max Moldavsky: 全方位共赢（all around positive impact） [11:46] Max Moldavsky: 知识与技能的民主化普及（democratization of knowledge and skills） [11:46] Max Moldavsky: keeping the plant full with minimum costs [11:46] Max Moldavsky: 从农田到餐桌（from farm to table） [14:01] Max Moldavsky: optimize every drop of data they have

Source quotes

[0:04] Max Moldavsky: age of less [0:04] Max Moldavsky: part of the problem [2:17] Max Moldavsky: If it's a good season, that's great. If it's a bad season, they're in trouble. We're in trouble. Everyone is in trouble. [4:36] Max Moldavsky: every drop of water [4:36] Max Moldavsky: every drop of data [4:36] Max Moldavsky: GrowSphere [9:17] Max Moldavsky: Crop Advisor [11:46] Max Moldavsky: democratization of knowledge and skills

Original transcript

[0:04] Hello everyone. Uh thanks for joining uh [0:07] the session and being here. My name is [0:09] Max Moldavski. I'm the director of [0:11] innovation and climate at Netaphim. Uh [0:14] Netim for those of you who are not [0:16] familiar with uh is the largest [0:19] irrigation company in the world but a [0:20] bit more on that later. Uh today I'm [0:23] going to tell you a bit about something [0:26] different from what we seen mostly in [0:28] the last three days here in the [0:30] convention. A bit on how AI is shaping [0:33] the next revolution in agriculture. [0:36] So [0:37] like in any good story let's start with [0:40] the problem and the problem is that we [0:42] are living in the age of less less what? [0:46] So first of all less water. We're seeing [0:48] less and less available fresh water uh [0:51] with every year that is going by. Uh the [0:54] water resources that we have as humanity [0:56] are shrinking and even in places where [0:59] you wouldn't expect it uh in uh you know [1:02] in a matter of climate and so on. Then [1:06] after water we have less arable land [1:08] that is used to grow our food. In the [1:10] last 40 years we lost onethird of the [1:13] global arable land that is fit for [1:15] agriculture. And last but not least, [1:18] less stability. In the last years, we're [1:21] seeing extreme weather events increasing [1:23] more and more, whether it's droughts, [1:25] floods, wildfires, storms, and so on [1:29] that are impacting severely what we uh [1:31] are able to do with our land. [1:35] And a bit paradoxically that in this age [1:37] of less, we have more more what? Well, [1:41] more mouth to feed. With the growing uh [1:44] population and the changes in the social [1:48] dynamics, we will need 60 to 70 more uh [1:52] food to feed everyone globally uh by the [1:55] by the year 2050 with as you just seen [1:58] less resources uh that we have to do [2:02] that. And that means that the way that [2:04] we grow our food today is not only not [2:08] sustainable for this goal but it's even [2:10] part of the problem. Um and let's see on [2:14] one example of why and let's focus on [2:17] how inefficiently we are using water and [2:20] land. So in terms of global usage of [2:23] fresh water and this is something that [2:26] not everyone intuitively knows but 70% [2:28] goes to agriculture. Everything that we [2:31] use in our home or in factories and so [2:34] on is only 30%. But this 70% of water [2:38] that goes to agriculture is used only to [2:41] uh irrigate 24% of the land that grows [2:43] our food. The rest the 76% of the other [2:47] land farmers are relying upon rain. If [2:51] it's a good season, that's great. If [2:53] it's a bad season, they're in pro. [2:55] They're in trouble. We're in trouble. [2:56] Everyone is in trouble. But the biggest [3:00] tra the biggest problem here is that [3:01] even in this 24% [3:04] 75% of these lands are irrigated in the [3:07] same way that our ancestors are using [3:11] used to irrigate their land hundreds and [3:13] thousands years ago in very inefficient [3:15] ways. [3:17] And that's where precision irrigation [3:19] comes to play. Precision irrigation is [3:22] the way to really optimize every drop of [3:24] water to get best results both in terms [3:27] of resources used and in terms of [3:30] results per a drop of water per piece of [3:33] land and so on. [3:36] So the advantages beyond the the almost [3:39] trivial savings of water that can be on [3:42] average 50% sometimes even more [3:44] depending on the crop is first and [3:46] foremost around yield. we are able to [3:49] produce 50 sometimes even 100% more [3:52] yield in the same uh plot of land um [3:56] when we are switching to uh advanced [3:59] precision irrigation. Add to that the [4:01] reduced cost of other inputs like labor, [4:03] like energy, like crop protection and [4:05] the improved sustainability that we will [4:07] touch upon a little bit soon and you [4:09] will see the full picture of precision [4:12] irrigation. [4:14] That's what we've been doing doing in [4:15] NetApp for the last six years. We were [4:18] established in 1965 out of necessity in [4:21] the desert of Israel where we just [4:22] didn't had enough water to grow food and [4:25] we since then grew to be a multinational [4:29] company that serves millions of farmers [4:32] globally till today. [4:36] But is this all of the great things that [4:38] we've seen is this enough? So you can [4:40] guess my answer. The answer is no. We [4:42] need to take another step forward and [4:44] this next step step forward is the [4:46] digital farming revolution where we will [4:49] be able to accept optimizing every drop [4:52] of water like we discussed so far to [4:54] optimize also every drop of data. [4:59] So digital farming, this is the [5:02] revolution that we're talking about [5:03] right now. Basically gives farmers the [5:06] ability to take all the data that's [5:08] happening both on and off the farm to [5:11] optimize how they do whatever they do, [5:15] how do they apply resources, whether [5:17] it's water or other resources to get to [5:19] the optimal decision making and of [5:21] course autonomous decision making on the [5:24] farm. [5:25] At Netim, we are doing it around the [5:28] operating system that we called the [5:30] growth sphere that basically enables the [5:32] farmer to do 360 from planning to [5:36] monitoring whatever is happening in his [5:38] field through different types of sensors [5:42] to getting uh realtime recommendations [5:46] based on all of this data to what he [5:48] needs to do in terms of irrigation and [5:50] fertilization and all the way to closing [5:53] the loop with operating the system [5:55] itself based on these recommendations. [5:59] The system is built uh with two [6:02] different layers. The first layer is the [6:04] hardware layer that is comprised from [6:07] first of all the actual hardware that is [6:10] used for irrigation in the field whether [6:12] it's the irrigation system filters [6:15] valves and so on and also a set of uh [6:19] different sensors [6:21] uh that are enabling to get a lot of [6:23] data from the soil from the plant [6:26] themselves from everything that is [6:28] happening uh around like weather and [6:31] this is all connected to a second layer [6:34] which is the AIdriven software layer [6:36] that is taking all of the data from the [6:38] field combined with external resources [6:41] like remote sensing images from [6:44] satellites, drones and so on and weather [6:47] and and other external resources [6:50] taking into our crop models which are [6:53] the models that are enabling real time [6:57] uh mixing of all of these data to get [6:59] basically give recommendations [7:02] um that are live and are experienced [7:05] what is actually happening. For example, [7:07] maybe 2 days ago the recommendation was [7:09] to irrigate today but yesterday night it [7:13] was raining so we don't need to do it [7:15] anymore and we can postpone it in 3 [7:17] days. So the key AI elements we have a [7:21] lot but the key AI elements in the [7:23] system first of all all of these models [7:25] are AI based. So there is a lot of [7:29] knowledge that was collected through [7:31] these 60 years of experience but it's [7:34] not always enough and you really need to [7:36] have much more ideas of what to actually [7:40] do with the data and the AI is enabling [7:42] us to enhance our aronomical knowledge. [7:46] Then the second element is the hyper [7:49] localization. [7:51] Aronomists usually give recommendations [7:53] that are I wouldn't say generic but are [7:56] relatively general and in the end every [7:59] plot of land every piece of soil is very [8:02] different. I can have my plot of land [8:04] and my neighbor which is a few hundred [8:09] of meters away from me and the soil will [8:11] be completely different. AI enable us to [8:14] really learn every specific piece of [8:16] plot to give best recommendation to it [8:19] specifically. [8:21] The third element is the agent is this [8:24] uh built in into the system. Farming is [8:28] difficult. Farming requires the farmer [8:30] to have a lot of knowledge whether it's [8:33] how to irrigate, when to seed, which [8:35] seeds to choose, when to apply crop [8:38] protection and so on so on. It's a [8:40] difficult job uh with a lot of knowledge [8:43] that is required from him and we want to [8:45] make it easier at least in our domain [8:48] for the farmer to speak with the system [8:51] understand easily what are the [8:52] recommendations understand what are [8:54] these different charts that we are [8:56] showing them and have the ability to see [8:59] the recommendations in a very simple way [9:02] and of course allow the agent to [9:04] actually automate the system in his [9:06] behalf if he chooses to do so. [9:10] So [9:11] after all of the we talked, let's see [9:13] how does it actually look from the [9:15] farmer's perspective with a short clip. [9:17] [Music] [9:19] >> Hey everyone. [9:21] So [9:23] how are we doing with our irrigation? [9:26] >> Uh plants are almost ready. Just need a [9:27] quick update cuz it's going to be hotter [9:29] than we thought. 3.4° higher to be [9:32] precise. [9:33] Yeah, some of the crops in pot 21 and [9:35] looking at their thirsty [9:38] >> and soil water content in pots 12, 15, [9:41] and 33 is too high. [9:43] >> I'm heading to the northern section. [9:45] I'll report back in about 5 hours. [9:47] >> This is a mess. [9:48] >> Hey, are we over or are we under [9:49] irrigating right now? [9:50] >> Under? [9:51] >> Definitely over. [9:53] >> Why does it have to be so complicated? [9:55] Actually, it doesn't. We can use Gross [9:57] OS. [10:00] >> What's that? Growphere is an operating [10:03] system for irrigation and fertigation [10:04] that helps us to plan, control, and [10:06] monitor everything all in one place. [10:09] Real how Growphere connects our [10:12] irrigation plans to the controller, to [10:14] the hydraulic system, to the field [10:16] sensors, giving us realtime irrigation [10:18] recommendations. [10:19] >> Okay, let's see what it can do. [10:22] >> All right, check it out. We can see the [10:24] irrigation status of all our plots at [10:26] any time [10:28] >> without having to drive out to the [10:30] field. Yeah, [10:31] >> awesome. Cuz I hate driving to the [10:34] northern section. [10:35] >> Hold on. I see that there's an alert on [10:36] plot 7, incomplete scheduled irrigation, [10:40] and I see the sensors are showing dry [10:42] soil. [10:42] >> No worries. Gross's already updated [10:44] tomorrow's irrigation schedule [10:46] accordingly. [10:46] >> But what if there's a sudden rainstorm? [10:48] >> Growhere automatically factors this into [10:50] our irrigation plans, too. [10:52] >> H, that's pretty smart. [10:54] >> What if we start growing a variety we [10:56] haven't grown before? Growier's Crop [10:58] Advisor can help us with that, too. [11:00] Using data from similar crops growing [11:02] around the world. And [11:03] >> you can do all of this from anywhere. [11:05] >> Anywhere. Your computer, tablet, even [11:08] your phone. [11:09] >> Fantastic. Cuz [11:10] >> you hate the Northern Sack Shang. So, [11:14] Growphere, huh? [11:16] >> Growhere. [11:17] >> Growhere. [11:19] [Music] [11:23] Love it. [11:30] So what one of my favorite experience [11:32] maybe in the last year was to speak with [11:34] a farmer that that told me look I had [11:38] one alert that enabled me to understand [11:41] that there's something wrong in one of [11:43] my plots [11:45] and there is zero chance that I would [11:46] have understood it without looking at [11:48] the system and really saved all of the [11:51] all of this plot and otherwise probably [11:53] I would have zero income from and this [11:55] is what we're trying to do and the idea [11:59] is that it enables everyone to win to [12:02] have all around positive impact first of [12:03] all for the plants and the planet. Think [12:06] of improving and maintaining soil health [12:08] because of smart and efficient use of [12:11] resources. Um think about the farmers [12:14] which obviously first of all they're [12:16] improving their yield and profitability [12:18] allowing them to grow [12:21] food for us in a a sustainable way for [12:24] them. But maybe even more importantly, [12:28] we we need to remember that the vast [12:30] majority of farmers today are small [12:33] holders. Farmers that are uh cultivating [12:36] on a single digit amount of acres [12:40] and they don't have the access or the [12:42] funds to have professional support from [12:46] aronomists on water, on fertilizers, on [12:49] crop protection, on the thousand things [12:50] that they need to do. And they're just [12:52] trying to do their best. Think of the [12:55] fact that with AI, with models that are [12:58] not depending on a person that will come [13:01] to help them, we are basically allowing [13:03] for this democratization of of knowledge [13:06] and skills to have everyone in the same [13:08] level and enabling everyone to get best [13:11] results. [13:12] Then the food industry is also a very [13:14] important player here. Food industry [13:16] cares a lot about ensuring a robust [13:18] supply chain uh and a predictable one at [13:21] that. uh their one one of their most [13:24] important KPIs is keeping the plant full [13:27] with minimum costs. So think of the fact [13:30] that their farmers having that and them [13:33] having this view of everything that is [13:35] going on with the farms that they're [13:36] procuring uh crops with that's a big [13:39] advantage to them. And last but not [13:42] least, think about the fact that it [13:44] enable enables full traceability from [13:48] farm to table. That helps us the [13:50] consumers in the end to understand [13:52] exactly what happened with the food that [13:55] we are eating all the way from the farm. [14:00] Now [14:01] we zoomed in a little bit on on on [14:04] specifically where does it come in [14:06] precision irrigation but there is a lot [14:09] more potential that it can come from uh [14:12] in terms of AI that is applied in [14:15] agriculture. [14:16] Think of things like advanced models for [14:19] the super important financial and [14:21] insurance institution that will enable [14:24] farmers to get much better financing or [14:27] insurance that they can't get today [14:29] because the we will have the ability to [14:31] create models that these institutions [14:34] will believe and be able to uh get [14:37] better uh premiums for the farmers or [14:40] even things that are today so far from [14:43] us. think of uh thinking of new ways to [14:46] to cultivate crops. Maybe the common [14:50] conception today that you need to plant [14:53] for example corn in March or April and [14:57] to and to harvest it in September, [15:00] October. Maybe it's wrong. Maybe we will [15:02] be able to find new models that are [15:04] again super hyper localized to get much [15:07] better results. So it's just a taste but [15:11] we are heading to a revolution that will [15:13] really put even the early 20th century [15:16] agricultural revolution which was huge [15:19] uh where we grew tfold the amount that [15:22] of food that we are growing in my view [15:25] it will make it even a bigger evolution. [15:30] So thanks thanks very much for uh [15:33] attending and listening. Um we are [15:36] really glad to lead this AI powered next [15:40] revolution in terms of food uh and [15:44] helping millions of farmers to optimize [15:47] again every drop of data they they have. [15:49] Thank you very much.