Netafim Pitches Farm Irrigation to Offset AI Data Center Water Use

AI Water Stewardship: Farms & Data Centers in Shared Basins With Netafim — Ai4

Ai4

Data centers cannot resolve shared-basin deficits through on-site cooling efficiency alone, requiring tech operators to fund agricultural precision irrigation upgrades to generate auditable replenishment credits and secure compute growth.

AI-assisted summary of AI Water Stewardship: Farms & Data Centers in Shared Basins With Netafim — original source. Check the original for full context.

Source published: 2026-08-17

Detailed notes

## Background Published on YouTube by the Ai4 channel on August 17, 2026, this presentation features Max Moldavsky, Director of Innovation and Climate Solutions at Netafim. The session outlines how AI-powered precision irrigation enables measurable basin-level water stewardship, harmonizing agricultural productivity with the escalating water demands of digital infrastructure.

## TL;DR The talk addresses the shared resource bottleneck facing AI infrastructure and global agriculture, both of which rely on finite freshwater supplies in localized watersheds. Because agriculture accounts for 70% of global freshwater consumption—mostly via inefficient flood irrigation—corporate funding of precision irrigation provides tech operators with massive, verifiable water replenishment credits and community trust. Netafim illustrates this approach through GrowSphere OS, an operating system that pairs IoT telemetry with hyper-local AI crop models and LLM agents to automate daily field fertigation. Ultimately, precision irrigation creates a closed loop where AI-driven predictive intelligence safeguards the very water basins required for digital and agricultural coexistence.

## Atmosphere & style The presentation maintains an authoritative, pragmatic, and strategic tone, approaching resource scarcity not with alarmism, but as a solvable operational engineering challenge. Moldavsky delivers his address with executive clarity, systematically unpacking the physical infrastructure behind cloud computing alongside boots-on-the-ground agronomic realities. It feels like an insightful corporate blueprint that seamlessly bridges high-level tech sustainability goals with tangible, field-level automation.

## Key takeaways - [00:04](https://www.youtube.com/watch?v=Po_UagChdlA&t=4s) Water stewardship for digital infrastructure is an operational resilience imperative and social license to operate rather than a public relations exercise. - [01:47](https://www.youtube.com/watch?v=Po_UagChdlA&t=107s) Shared water basins face compounding strain from exponential AI compute cooling demands and the need to produce 60% to 70% more food by 2050. - [04:35](https://www.youtube.com/watch?v=Po_UagChdlA&t=275s) Internal data center water efficiency improvements reach physical limits, making basin-wide agricultural stewardship the primary leverage point for watershed resilience. - [04:35](https://www.youtube.com/watch?v=Po_UagChdlA&t=275s) Agriculture accounts for roughly 70% of global freshwater consumption, with 75% of irrigated land still relying on inefficient flood irrigation. - [07:26](https://www.youtube.com/watch?v=Po_UagChdlA&t=446s) Transitioning legacy flood-irrigated farmland to precision root-zone delivery can reduce water application by 50% or more while boosting crop yields. - [07:26](https://www.youtube.com/watch?v=Po_UagChdlA&t=446s) Corporate water stewardship partnerships with Netafim follow an empirical, verified pathway to deliver auditable volumetric replenishment credits. - [09:24](https://www.youtube.com/watch?v=Po_UagChdlA&t=564s) Basin-level precision irrigation generates multi-faceted co-benefits, including agricultural greenhouse gas mitigation and improved farmer economic viability. - [11:52](https://www.youtube.com/watch?v=Po_UagChdlA&t=712s) AI functions as both a driver of water demand in data centers and an optimizing intelligence layer across agricultural operations. - [13:23](https://www.youtube.com/watch?v=Po_UagChdlA&t=803s) GrowSphere OS combines field IoT telemetry, hydraulic controls, and hyper-local machine learning crop models to automate daily fertigation schedules. - [13:23](https://www.youtube.com/watch?v=Po_UagChdlA&t=803s) LLM-based farm agents help eliminate farmer decision fatigue by translating complex agronomic conditions into automated field actions. - [17:45](https://www.youtube.com/watch?v=Po_UagChdlA&t=1065s) Pairing predictive AI with physical precision irrigation allows sustainable digital infrastructure growth and agricultural productivity to coexist in shared basins.

## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=Po_UagChdlA&t=4s) The Unseen Link Between AI, Food, and Water Resilience

Artificial intelligence, global food production, and agriculture are deeply interconnected around a single critical resource: freshwater. Max Moldavsky emphasizes that tackling this intersection under the banner of "Shared water and shared future" is neither a public relations exercise, a marketing campaign, nor mere corporate messaging. Instead, it is an imperative centered on operational resilience—ensuring organizations maintain business continuity without disruption, build long-term trust with surrounding communities, and secure the social "license to grow the business."

While AI stands as the next transformative technological revolution, it is underpinned by a heavy physical infrastructure layer. Surging, exponential demand for compute capacity translates directly into escalating requirements for power and energy, which in turn demands massive volumes of water for data center cooling and continuous operation.

[01:47](https://www.youtube.com/watch?v=Po_UagChdlA&t=107s) Shared Basins: Coexisting Demands of AI and Global Food Supply

AI and agriculture may appear to operate in completely separate realms, yet both are foundational systems that rely on the exact same localized natural resources. Water is finite and increasingly diminishing, facing heightened competition across agricultural production, digital infrastructure operations, and municipal consumption.

```mermaid flowchart LR accTitle: Competing Pressures on Shared Basin Water Resources subgraph Drivers ["Growing Regional Pressures"] A["Exponential AI Compute Demand"] B["60-70% More Food Needed by 2050"] end subgraph Bottleneck ["Shared Basin Constraint"] C["Finite & Diminishing Freshwater Resources"] end subgraph Solutions ["Strategic Imperatives"] D["On-Site Data Center Efficiency"] E["Basin-Level Precision Stewardship"] end A --> C B --> C C --> D C --> E ```

This resource crunch is exacerbated by global population growth. By 2050, the world will require 60% to 70% more food than is produced today, even as available freshwater reserves decline. Because of this reality, AI infrastructure can no longer be treated as an abstract, purely digital domain; it is inherently bound to local watershed dynamics. When a data center is built, local stakeholders—including farmers and regional residents—inevitably confront pressing questions about basin capacity: - Will there be sufficient water in the local basin to sustain the community? - Can rapid AI expansion coexist with the agricultural expansion required to feed the population? - How can technology operators demonstrate authentic, responsible resource stewardship?

Simultaneously, technology companies face the existential operational question of whether regional water supplies will remain sufficient to prevent business discontinuation and support future capacity expansion. Navigating this constraint makes being perceived as a reliable, constructive "good neighbor" in local basins a primary competitive advantage for tech infrastructure providers.

[04:35](https://www.youtube.com/watch?v=Po_UagChdlA&t=275s) Beyond Data Center Efficiency: The 70% Agricultural Leverage Point

Maximizing internal water efficiency within data center cooling systems is mandatory, yet on-site optimizations have physical limits and cannot solve basin-wide deficits on their own. True water stewardship requires looking beyond the facility boundary to fund projects that create verified, large-scale water savings across the shared watershed.

While initiatives such as rainwater harvesting, graywater recycling, and distribution infrastructure upgrades offer value, agriculture presents the single highest leverage point for water conservation.

- **Global freshwater consumption:** Agriculture accounts for approximately 70% of all freshwater consumed worldwide. - **Prevalence of legacy methods:** Roughly 75% of global irrigated land still relies on traditional flood irrigation—a centuries-old practice that inundates entire fields, resulting in heavy runoff and evaporation losses. - **Precision irrigation adoption:** Modern precision irrigation accounts for only 7% of irrigated farmland globally. - **Conservation potential:** Transitioning from flood to precision irrigation can reduce agricultural water application by 50% or more.

[07:26](https://www.youtube.com/watch?v=Po_UagChdlA&t=446s) Delivering Water to the Root Zone: Precision Irrigation Stewardship Programs

Precision irrigation fundamentally transforms water delivery by directing moisture and nutrients directly to the root zone of the crop rather than broadcasting it across the soil surface. This targeted delivery enables growers to "grow more with less," producing significantly higher crop yields while simultaneously reducing input costs for synthetic fertilizers, crop protection chemicals, labor, and pumping energy.

```mermaid flowchart LR accTitle: Corporate Precision Irrigation Stewardship Workflow subgraph Corporate ["Corporate Footprint"] A["Identify Target Basin"] --> B["Calculate Replenishment Volume"] end subgraph Implementation ["Netafim Program Execution"] B --> C["Enroll Legacy Flood Farmers"] C --> D["Establish Empirical Baseline"] D --> E["Deploy Precision Hardware & IoT"] end subgraph Outcomes ["Shared Basin Impact"] E --> F["Auditable Water Reduction"] F --> G["Drought Resilience & Growth License"] end ```

For corporate enterprises seeking water replenishment credits, funding agricultural precision irrigation delivers an exceptionally high return on investment by generating massive, auditable water savings in the basins where they operate. Netafim structures these watershed programs through a multi-step execution model: 1. **Basin identification:** The corporate partner identifies a priority basin where it currently operates or plans to expand digital infrastructure. 2. **Volumetric modeling:** The required replenishment volume is calculated to achieve water-neutral or water-positive performance targets. 3. **Grower enrollment:** Netafim recruits local farmers currently using inefficient flood irrigation methods and provides the funding and equipment needed to upgrade their fields. 4. **Empirical baselining:** Precise historical and current water consumption baselines are established using field data rather than top-down estimations. 5. **Implementation & verification:** High-efficiency irrigation and continuous IoT monitoring systems are installed, generating verifiable, year-over-year volumetric water reductions.

[09:24](https://www.youtube.com/watch?v=Po_UagChdlA&t=564s) Verifiable Water Replenishment and Systemic Co-Benefits

These stewardship programs provide tech companies with credible, verified progress toward their corporate water-positive commitments. Beyond satisfying reporting targets, the programs materially reduce exposure to regional drought risks and regulatory scrutiny, ensuring data centers can operate smoothly and expand compute capacity without threatening local water security. Investing directly in local farm productivity builds genuine community trust, reframing digital infrastructure as a collaborative partner in a shared future.

The positive impacts extend beyond volumetric water conservation: - **Greenhouse gas mitigation:** Precision irrigation significantly reduces agricultural greenhouse gas emissions by preventing anaerobic soil conditions caused by over-flooding and optimizing fertilizer application. - **Farmer economic resilience:** Higher crop yields and lowered operational expenses improve long-term farm viability. - **Basin relief:** Reduced aggregate agricultural withdrawals alleviate systemic pressure on shared municipal and ecological water supplies.

[11:52](https://www.youtube.com/watch?v=Po_UagChdlA&t=712s) GrowSphere: Closing the Loop with AI and Farm Operating Systems

AI plays a dual role in this ecosystem—both as a driver of water demand in data centers and as an optimizing force in the field. Netafim integrates AI directly into precision agriculture through its GrowSphere operating system, which establishes an automated, closed-loop management environment for growers.

GrowSphere unifies automated field execution, continuous agronomic and hydraulic monitoring, and real-time operational recommendations to deliver daily, actionable insights for farmers. The system operates across a two-layer architecture: - **Physical field infrastructure:** Driplines, automated valves, filtration assemblies, and pumps configured for precise water and nutrient delivery. - **Sensory and control telemetry:** In-field agronomic sensors (measuring soil moisture and crop status) and hydraulic monitoring devices (tracking line pressures, flow rates, and distribution anomalies) connected directly to smart controllers.

[13:23](https://www.youtube.com/watch?v=Po_UagChdlA&t=803s) Hyper-Local AI Crop Models and LLM Agents for Farm Management

At the software and cloud layer, precision irrigation systems integrate multiple external data streams—including real-time weather forecasts, satellite imagery, and drone footage—directly into crop modeling algorithms. Unlike traditional static analytical models, these systems are powered by machine learning architectures that continuously evolve and refine their recommendations season after season. A core imperative behind this design is handling hyper-locality: no two agricultural plots are identical, even when they sit directly adjacent to one another. Differences in historical land use, previous crop rotations, tillage methods, and soil treatments mean that simple macro-rules—such as "it rained yesterday, so do not irrigate today"—fall short of effective resource management.

```mermaid flowchart TD accTitle: Precision Irrigation Cloud and AI Decision Flow subgraph Inputs ["Data Inputs"] S["Field & Soil Sensors"] W["Weather Forecasts"] I["Satellites & Drones"] end

subgraph Platform ["GrowSphere & AI Engine"] M["Hyper-Local Crop Models"] A["LLM-Based Farm Agents"] end

subgraph Actions ["Field Execution"] C["Hydraulic Controllers"] R["Optimized Daily Fertigation"] end

S --> M W --> M I --> M M --> A A --> C C --> R ```

Agriculture places an immense cognitive and operational load on farmers, who must simultaneously serve as experts in irrigation scheduling, crop protection chemistries, seed variety selection, fertilization timing, and harvesting logistics. To alleviate this decision fatigue and automate execution, Netafim incorporates an LLM-based agent architecture into the platform. This agent interprets real-time field data, translates complex agronomic conditions into clear insights, and enables farmers to automate routine field operations rather than managing every mechanical lever manually.

- **Continuous Model Training:** AI crop models evolve across cropping cycles to fine-tune daily irrigation and fertigation recommendations based on observed field conditions. - **Hyper-Local Variation:** Customizes inputs to individual plot histories and soil profiles rather than applying generalized regional assumptions. - **LLM Agent Assistance:** Empowers growers with conversational operational intelligence and automated field actions.

---

[15:34](https://www.youtube.com/watch?v=Po_UagChdlA&t=934s) GrowSphere OS: End-to-End Irrigation and Fertigation Automation

To illustrate how these digital tools function in everyday farm operations, a scenario demonstrates the confusion that arises during sudden environmental shifts. A farm team assesses upcoming weather and realizes temperatures will spike 3.4° higher than anticipated. Crop stress appears unevenly across the operation: crops in plot 21 show signs of water deficit, while soil moisture levels in plots 12, 15, and 33 remain excessively high. Facing the prospect of a five-hour drive just to manually inspect the northern section, the farm crew is deadlocked over whether they are currently "definitely over" or "under" irrigating.

``` +--------------------------+-------------------------------------------------------------+ | Operational Challenge | GrowSphere OS Automated Solution | +--------------------------+-------------------------------------------------------------+ | Unexpected Heat Spikes | Adjusts daily delivery volumes to offset high temperatures | +--------------------------+-------------------------------------------------------------+ | Remote Plot Monitoring | Cloud dashboard replaces 5-hour physical field visits | +--------------------------+-------------------------------------------------------------+ | Incomplete Irrigation | Flags dry soil alerts and recalibrates tomorrow's schedule | +--------------------------+-------------------------------------------------------------+ | Sudden Storms / Rain | Automatically credits precipitation into irrigation plans | +--------------------------+-------------------------------------------------------------+ | Unfamiliar Crop Types | Crop Advisor benchmarks data from similar worldwide crops | +--------------------------+-------------------------------------------------------------+ ```

The solution presented is GrowSphere OS, an operating system for irrigation and fertigation designed to plan, monitor, and execute field operations from a centralized interface. The software links irrigation schedules directly to hydraulic controllers, field-level soil sensors, and external forecasts, allowing farm operators to check status across every plot via smartphone, tablet, or PC without driving across the property. When an alert signals an incomplete irrigation cycle on plot seven alongside dry soil readings, the system automatically schedules compensatory watering for the following day. If an unexpected rainstorm hits, the platform recalibrates plans in real time. For growers introducing unfamiliar crops, a global crop advisor feature leverages datasets from comparable crops grown worldwide to generate baseline recommendations.

---

[17:45](https://www.youtube.com/watch?v=Po_UagChdlA&t=1065s) Closing the Loop: AI, Precision Irrigation, and Shared Basin Water Stewardship

In his closing remarks, Max Moldavsky emphasizes that water management must become a fundamental pillar of the broader AI and digital infrastructure conversation. As AI drives global computing demand and economic growth, precision irrigation provides the concrete physical mechanism to carry out measurable water stewardship on the ground.

By transitioning agricultural irrigation to precision, data-driven systems, water savings can be generated and verified at scale. This creates a cohesive closed loop between digital technology, food production, and agricultural sustainability, ensuring that data center expansion and agricultural operations can equitably and sustainably co-exist within the same shared river basins and aquifers.

- "Water is now an essential part of the conversation of AI that we need to have." - AI delivers the predictive intelligence, precision irrigation executes the physical conservation, and water stewardship aligns agricultural productivity with digital growth in shared basins.

## Notable quotes > Water is becoming the defining constraint for both food production and digital infrastructure—so how do we manage it smarter, together? — Max Moldavsky

> [00:04](https://www.youtube.com/watch?v=Po_UagChdlA&t=4s) Shared water and shared future — Max Moldavsky

> [00:04](https://www.youtube.com/watch?v=Po_UagChdlA&t=4s) license to grow the business — Max Moldavsky

> [01:47](https://www.youtube.com/watch?v=Po_UagChdlA&t=107s) Will there be sufficient water in the local basin to sustain the community? — Max Moldavsky

> [01:47](https://www.youtube.com/watch?v=Po_UagChdlA&t=107s) Can rapid AI expansion coexist with the agricultural expansion required to feed the population? — Max Moldavsky

> [01:47](https://www.youtube.com/watch?v=Po_UagChdlA&t=107s) How can technology operators demonstrate authentic, responsible resource stewardship? — Max Moldavsky

> [01:47](https://www.youtube.com/watch?v=Po_UagChdlA&t=107s) good neighbor — Max Moldavsky

> [07:26](https://www.youtube.com/watch?v=Po_UagChdlA&t=446s) grow more with less — Max Moldavsky

> [13:23](https://www.youtube.com/watch?v=Po_UagChdlA&t=803s) it rained yesterday, so do not irrigate today — Max Moldavsky

> [17:45](https://www.youtube.com/watch?v=Po_UagChdlA&t=1065s) Water is now an essential part of the conversation of AI that we need to have. — Max Moldavsky

## Mentioned Max Moldavsky, Netafim, Ai4, GrowSphere, GrowSphere OS

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

Water is becoming the defining constraint for both food production and digital infrastructure—so how do we manage it smarter, together? In this talk, I will explain how precision irrigation, powered by AI, turns water stewardship into measurable action at the basin level. By combining sensors, forecasts, and crop models, we can optimize every drop—reducing withdrawals, improving yields, and generating verifiable water savings. I will show how these auditable outcomes support corporate targets while strengthening shared-basin resilience. Join this session to explore how data-driven precision irrigation can align agricultural productivity with sustainable digital growth.

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

中文简报

Netafim提议通过农田灌溉抵消AI数据中心用水数据中心仅凭自身的冷却效率无法解决共享流域的水资源短缺问题，这要求科技运营商出资支持农业精准灌溉升级，以产生可审计的补水额度并保障算力增长。本次演讲探讨了AI基础设施与全球农业共同面临的资源瓶颈，两者均依赖局部流域内有限的淡水供应。鉴于农业占全球淡水消耗量的70%——且大部分仍采用低效的大水漫灌——企业出资支持精准灌溉能够为科技运营商提供大规模、可验证的补水信用额度以及社区信任。Netafim通过GrowSphere OS阐释了这一方案，该操作系统将IoT遥测技术与超本地AI作物模型及LLM智能体相结合，以实现农田日常水肥一体化的自动化。最终，精准灌溉形成了一个闭环，利用AI驱动的预测智能来保护数字与农业共存所必需的水资源流域。数字基础设施的水资源管理是一项关乎业务运营韧性与获取社会许可的根本要求，而非单纯的公关宣传或营销活动。 (0:04) 随着 AI 计算冷却需求爆发式增长，以及到 2050 年全球需增产 60% 至 70% 粮食的压力，共享流域正面临日益严峻的水资源瓶颈。 (1:47) 农业占全球淡水消耗的约 70%，且 75% 的灌溉农田仍采用漫灌；将传统漫灌升级为精准根区灌溉可减少 50% 以上的农业用水并提高作物产量。 (4:35) 企业与 Netafim 合作开展精准灌溉水资源管理项目，通过经验证的路径获得可审计的水资源回补指标，同时带来农业温室气体减排与农户增收等协同效益。 (7:26) Netafim 依托 GrowSphere 系统引入超本地化 AI 作物模型与 LLM 智能体，在精准农业中实现灌溉与施肥全流程自动化管理，形成闭环优化。 (11:52)

Reading summary

Max Moldavsky argues that expanding AI data infrastructure and global food systems rely on the same finite freshwater basins, making corporate investment in agricultural precision irrigation the highest-leverage mechanism to achieve verified water replenishment and preserve operational continuity. The Shared Basin Resource Dilemma (1:47) Artificial intelligence relies on physical infrastructure that requires substantial electricity and water cooling, placing data centers in direct competition with local communities and agriculture for finite freshwater reserves. Moldavsky stresses that addressing this resource intersection is not a branding exercise, but an operational necessity to maintain business continuity and retain community trust. This resource strain is intensifying as global population growth is projected to require 60% to 70% more food production by 2050, even as freshwater reserves dwindle. Technology providers expanding their compute footprint face existential operational risks if regional basins run short of water, transforming local environmental stewardship into a core competitive requirement. Agricultural Efficiency as a Watershed Leverage Point (4:35) While internal data center cooling efficiencies are necessary, on-site mechanical improvements have physical thresholds and cannot remediate basin-wide deficits on their own. Agriculture represents the single largest leverage point for watershed-scale conservation, accounting for roughly 70% of global freshwater withdrawals. Approximately 75% of global irrigated acreage still relies on legacy flood irrigation, which causes extensive runoff and evaporation losses, while modern precision irrigation covers only 7% of irrigated land. Transitioning fields from flood systems to precision root-zone delivery can reduce agricultural water application by 50% or more while increasing crop yields and cutting fertilizer, chemical, and pumping expenses. Corporate Replenishment Workflows and Co-Benefits (7:26) For technology firms pursuing water-positive commitments, funding agricultural precision irrigation offers auditable replenishment credits within the specific basins where they operate. Netafim's corporate workflow models required replenishment volumes, recruits local flood-irrigating growers, establishes empirical field baselines, and installs precision hardware alongside IoT monitoring to document year-over-year volumetric water reductions. These interventions provide system-level co-benefits beyond data center risk mitigation. Precision delivery curtails agricultural greenhouse gas emissions by preventing the anaerobic soil conditions caused by field flooding, enhances farm profitability, and eases pressure on municipal and ecological supplies across the shared watershed. AI and Operating Systems in the Field (11:52) AI acts simultaneously as a consumer of water in data centers and an efficiency driver in agriculture through platforms like Netafim's GrowSphere OS. Because hyper-local variables like tillage history and soil variation prevent broad macro-rules from working effectively, machine learning models continuously process drone imagery, satellite inputs, weather forecasts, and in-field soil sensors to fine-tune daily irrigation. To counter operational complexity and farmer decision fatigue, the platform deploys an LLM-based agent to interpret agronomic conditions and automate field actions. The operating system links software directly to hydraulic valves and pumps, dynamically compensating for unpredicted heat spikes, crediting unexpected rainfall, and flagging field anomalies from a centralized cloud interface. Moldavsky concludes that freshwater management must become a fundamental pillar of digital infrastructure strategy. Deploying AI-assisted precision irrigation creates a closed-loop framework where technological expansion actively supports agricultural productivity, enabling compute infrastructure and farming to sustainably coexist in shared river basins.

精炼阅读摘要

Max Moldavsky指出，人工智能的算力爆发与全球粮食增产对有限淡水资源的共同依赖，已将水资源管理推向企业运营韧性的核心；科技企业仅依靠优化数据中心内部冷却无法化解流域赤字，唯有资助将高耗水传统漫灌升级为农业精准灌溉，才能在共享流域中建立可审计的节水信用与长远发展许可。共享流域下算力扩张与粮食安全的冲突 (1:47) 随着人工智能计算需求呈指数级增长，底层实体基础设施对电力和冷却用水的需求急剧上升，使数字基建与农业生产在局部流域内产生了直接的资源竞争。Max Moldavsky强调，应对水资源紧缺绝非单纯的企业公关或品牌营销，而是关乎数据中心业务连续性与防范运营中断的生存底线。科技运营商唯有被周边社区与农户视为负责任的合作者，才能在面临流域水资源紧张时确立社会许可与长期的扩张资格。与此同时，全球人口增长推动粮食需求持续攀升。根据预测，到2050年全球所需粮食将比目前增加60%至70%，但可用的淡水储备却在日益减少。这一现实意味着人工智能基础设施不能被割裂为纯粹的虚拟算力，其物理命脉完全受制于所在地含水层和河流动态。当农业扩张与算力扩张在有限的自然水文循环中狭路相逢，水资源承载力成为了限制数字经济发展的决定性边界。超越机房局限：农业精准灌溉的高杠杆 (4:35) 尽管持续提升数据中心内部冷却系统的水资源利用效率是行业必须履行的基本标准，但设施边界内的技术优化存在明显的物理天花板，单凭机房节水根本无法填补整个流域的生态赤字。真正的水资源管理必须向外延伸，在共享流域内寻求规模化且经过核实的真实节水突破口。虽然雨水收集和中水回用具备一定补充价值，但全球农业才是杠杆率最高的节水切入点。统计数据表明，农业消耗了全球约70%的淡水资源。然而，在目前全球受灌溉的农田中，仍有约75%沿用大水漫灌这一具有数百年历史的传统模式，造成严重的地表径流和无效蒸发损失；与之相对，现代精准灌溉技术的全球采纳率仅为7%。如果将农田从大水漫灌转向直接输送水肥至作物根系的精准灌溉，田间水资源施用量可减少50%甚至更多，为紧张的流域水网释放出巨大的回旋空间。企业水资源回补模式与多维协同效益 (7:26) 针对科技巨头设立的“水资源净正效益”（water-positive）目标，Netafim提供了一套结构化的流域执行流程：科技企业首先锁定拟运营或扩张的重点流域，测算达成中和所需的水回补总量；随后，项目团队在当地招募仍在使用大水漫灌的农户，为其安装精准灌溉硬件与物联网监控设备；最终基于实地监测的历史与现状基线，生成经得起第三方审计的年度节水量，用于抵消数据中心的水足迹。这种企业注资农业转型的机制不仅能切实降低数据中心面临的干旱风险与监管问责，还带来了显著的系统性协同效益。在生态层面，精准灌溉避免了过度淹水引发的土壤厌氧环境，并通过优化水肥一体化方案，大幅削减了农业温室气体排放；在经济层面，定向输送大幅降低了农户在化肥、农药、抽水电力和人工上的投入，同时显著提升了作物单产，有效增强了流域农业社区的经济抗风险能力与长期信任度。双重角色：以AI与操作系统赋能田间节水 (11:52) 在整个流域生态中，人工智能呈现出双重角色：它既是造成数据中心高水耗的驱动者，也是在田间实现极端节水的重要赋能者。Netafim通过其GrowSphere操作系统，将底层的滴灌管道、智能阀门、泵站与田间的土壤墒情传感器及水力监控设备无缝集成，构建出自动化、闭环式的田间管理体系，使过去粗放的灌溉调度全面升级为基于数据反馈的精准作业。在软件决策层，系统整合了气象预报、卫星影像和无人机多光谱数据，利用机器学习驱动的超本地化作物模型，消除因微气候和历史耕作差异造成的决策盲区。此外，系统内置的LLM智能体能够直接解读复杂的农艺数据并自动执行阀门调度，有效化解了农户在水肥配比与灌溉时机上的认知负荷。当出现气温意外飙升或个别田块灌溉未达标时，云端平台能自动重新校准次日灌溉计划，使农户无需耗费数小时驱车巡视即可精准化解干旱与过度浇灌风险。 Max Moldavsky总结认为，水资源管理必须成为人工智能基础设施讨论的核心组成部分。AI为农田提供了预测智能，而精准灌溉则在物理空间完成了可量化的节水闭环，二者的结合为科技算力与农业生产在共享流域的共存提供了可行路径；但该机制能否在全球更大范围内铺开，仍取决于科技资本与传统农业之间长期回补契约的建立速度，以及分散农户对数字化水肥工具的持续接纳度。

Detailed notes — zh

## Background 该视频由频道 Ai4 于 2026 年 8 月 17 日发布在 YouTube 平台。耐特菲姆（Netafim）创新与气候解决方案总监 Max Moldavsky 在演讲中探讨了水资源如何成为粮食生产与数字基础设施的共同瓶颈，并介绍了结合传感器、气象预测与作物模型的 AI 精准灌溉技术如何将水资源管护转化为流域层面的可量化行动。 ## TL;DR 随着 AI 算力基础设施与数据中心冷却用水需求的激增，科技企业与农业在共享流域内面临着日益严峻的资源竞争。由于数据中心内部节水存在物理极限，企业通过投资农业精准灌溉替代传统漫灌，不仅能在田间释放 50% 以上的节水潜力并获取经过审计核验的“水正效益”指标，还能推动农户增产增收。耐特菲姆推出的 GrowSphere 软硬件一体化系统，依托超本地化 AI 作物模型与内置 LLM Agent，实现了全天候监测、自适应灌溉调配与故障自愈执行。这一方案成功在 AI 算力扩张、粮食安全保障与流域生态韧性之间构建起互利共赢的闭环生态。 ## Atmosphere & style 整场演讲风格沉稳务实且极具前瞻性，主讲人跳出泛泛的公关叙事，直面算力狂飙背后的物理资源硬约束，论述逻辑环环相扣。现场穿插的农场管理痛点短片生动接地气，与云端 AI 作物模型和液压管网等硬核技术架构形成了鲜明呼应。整体氛围专业严谨，既有宏观产业协同的战略高度，又扎根于田间地头的真实泥土，传递出科技与自然共生破局的坚定力量。 ## Key takeaways - [00:04](https://www.youtube.com/watch?v=Po_UagChdlA&t=4s) 水资源管理已成为 AI 基础设施扩张的核心瓶颈与业务韧性支柱，直接关乎社区信任与业务扩张许可（License to grow），而非表面公关宣传。 - [01:47](https://www.youtube.com/watch?v=Po_UagChdlA&t=107s) 数据中心算力冷却需求与全球粮食增产需求在有限且递减的本地共享流域中形成直接资源竞争。 - [04:35](https://www.youtube.com/watch?v=Po_UagChdlA&t=275s) 数据中心内部节水存在物理极限，科技企业必须通过投资流域内的外部水资源管护（Water Stewardship）项目实现水回补与中和。 - [04:35](https://www.youtube.com/watch?v=Po_UagChdlA&t=275s) 农业消耗全球约 70% 的淡水，且 75% 的灌溉农田仍采用大水漫灌，精准灌溉具有 50% 以上的节水杠杆潜力。 - [07:26](https://www.youtube.com/watch?v=Po_UagChdlA&t=446s) 精准灌溉通过向作物根系靶向供给水肥，实现节水、增产、减药及降低能耗的多重协同效益（Grow more with less）。 - [08:55](https://www.youtube.com/watch?v=Po_UagChdlA&t=535s) 标准水资源管护项目通过现场传感器建立实测基准线，以经审计验证的真实节水量为企业提供可信赖的水正效益（Water positive）资产。 - [11:51](https://www.youtube.com/watch?v=Po_UagChdlA&t=711s) 耐特菲姆推出的 GrowSphere 系统构筑了软硬件一体化的农场操作系统，实现从田间感知、自动控制到动态建议的闭环管理。 - [13:23](https://www.youtube.com/watch?v=Po_UagChdlA&t=803s) 云端作物模型具备跨季度持续进化的超本地化（Hyper-local）洞察能力，能精准适应相邻地块因历史耕作带来的土壤微观差异。 - [14:41](https://www.youtube.com/watch?v=Po_UagChdlA&t=881s) 系统内置 LLM Agent 充当数字农艺助理，通过自然语言解析传感器警报并自动化执行决策，大幅减轻农户的繁复认知负荷。 - [15:25](https://www.youtube.com/watch?v=Po_UagChdlA&t=925s) GrowSphere OS 具备全地块远程监控、天气突变自动补水折算以及计划未完成时的闭环自愈调度能力。 - [17:45](https://www.youtube.com/watch?v=Po_UagChdlA&t=1065s) AI 提供了算力与决策智能，精准灌溉将其转化为节水行动，共同闭合了 AI 算力、粮食安全与流域共享未来的生态闭环。 ## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=Po_UagChdlA&t=4s) AI 基础设施与水资源的紧密连接：从公关叙事到业务韧性 AI、粮食与农业看似分属不同的领域，但它们之间存在着一种深刻却常被忽视的紧密联系，这一纽带正是水资源，即所谓的“共享之水，共享未来”（Shared water and shared future）。推动这一议题的核心绝非单纯的公关宣传（PR）、市场营销或品牌口号，而是关乎企业的实际“韧性”（Resilience）——使企业在业务连续性不受威胁的前提下持续运营，赢得周边社区的信任，并获得至关重要的“业务扩张许可”（License to grow）。随着 AI 革命的加速演进，算力需求正呈现指数级增长。然而，AI 并非纯粹的虚拟概念，它拥有庞大且不可忽视的物理基础设施底座。算力的扩张直接拉动了电力与能源消耗，而数据中心在运行与冷却过程中对水资源的需求也在急剧上升，水资源由此成为 AI 产业不可回避的核心要素。 - 核心驱动力：AI 算力基础设施的物理扩展需要大量能源与冷却水支持。 - 企业底层诉求：水资源管理关乎业务连续性、社区信任与扩张许可，而非表面化的公关动作。 --- ### [01:47](https://www.youtube.com/watch?v=Po_UagChdlA&t=107s) 共享流域的资源竞争：粮食增长与数据中心扩张的冲突农业与 AI 尽管形态各异，但它们具有一个根本共性：两者都是人类社会繁荣与运转所不可或缺的基础，且都极度依赖同一个地方性自然资源——水。水不仅是维持人类生存与农业生产的基石，也是维系数据中心运转的命脉。然而，水资源不仅储量有限，且在全球范围内正面临不断减少的严峻现实。与水资源短缺形成鲜明对比的是需求的双重爆发：一方面是 AI 基础设施的几何级扩张，另一方面是全球人口增长带来的粮食缺口。预计到 2050 年，全球粮食需求将比现在增加 60% 至 70%。这使得数据中心无法再将自身视为孤立的数字实体，其建设与运营必然会触碰当地资源的敏感神经。在新建 AI 基础设施的地区，农民、居民等本地利益相关方已开始提出尖锐的质询：本地流域的水资源是否足以支撑社区生存？AI 的激增能否与农业增产共存？科技企业又该如何展现负责任的态度？对于科技巨头而言，确保在不发生断水风险的前提下持续扩容，已成为核心竞争力之一，他们必须学会在所在流域扮演“好邻居”（Good neighbor）的角色。 ```mermaid flowchart TD accTitle: AI 与农业在共享流域中的资源竞争与协同 subgraph S1["本地流域有限水资源 (Shared Basin Water)"] W["有限且递减的淡水供应"] end subgraph S2["双重增长需求"] AI["AI 算力基础设施<br/>(数据中心冷却与电力需求)"] AG["农业与粮食生产<br/>(2050年粮食需求增加60%-70%)"] end subgraph S3["核心挑战与诉求"] C1["科技企业: 业务连续性与扩张许可"] C2["本地社区与农户: 水安全与生存生计"] end W --> AI W --> AG AI --> C1 AG --> C2 C1 <-->|"水资源协同管理 (Water Stewardship)"| C2 ``` --- ### [04:35](https://www.youtube.com/watch?v=Po_UagChdlA&t=275s) 超越数据中心自身节水：为何聚焦农业精准灌溉面对用水压力，仅仅提高数据中心自身的冷却水利用效率是远远不够的，这类内部工程优化存在物理极限。企业必须将视野扩展到“水资源管理”（Water Stewardship），通过出资支持能够在当地流域产生可量化、可验证节水效果的外部项目，从而弥合水资源缺口。在雨水收集、中水回用、管网升级等诸多节水路径中，潜力最大、最具杠杆效应的领域是农业，特别是精准灌溉（Precision Irrigation）。数据显示，尽管全球工业与技术飞速发展，农业用水依然占据了全球淡水消耗总量的 70% 左右。更具改善空间的是，目前全球约 75% 的灌溉农田仍在使用几千年来沿袭的大水漫灌（Flood Irrigation），这种方式导致大量水资源在漫流和土壤下渗中浪费；而高效的精准灌溉技术在全球的普及率仅有 7% 左右。通过技术替代，精准灌溉能够在田间实现 50% 甚至更高的节水幅度，这是任何其他单项节水措施都无法比拟的巨大潜力空间。 - 全球淡水消耗占比：农业占约 70%。 - 传统灌溉现状：全球约 75% 的灌溉土地仍采用效率低下的大水漫灌。 - 精准灌溉普及率：目前全球仅覆盖约 7% 的灌溉农田。 - 节水潜力：相比传统漫灌方式，精准灌溉可节省 50% 以上的用水量。 --- ### [07:26](https://www.youtube.com/watch?v=Po_UagChdlA&t=446s) 精准灌溉的价值模型：“以更少资源产出更多” 精准灌溉的本质是将水和养分直接精准输送至作物的“根系区域”（Root zone），而不是随意洒在土壤表面。这种直接作用于根部的机制带来了全方位的协同效益：除了直接节省大量水资源外，还能显著提高农作物的单产与品质，并减少化肥、农药、劳动力和电力等农业投入品的消耗。耐特菲姆将其核心价值主张概括为“以更少资源产出更多”（Grow more with less）。在水资源管护模式下，精准灌溉展现出极高的投资回报率（ROI）。科技企业通过出资为当地农户配备先进的灌溉技术，既能帮助农户实现生产升级，又能为企业自身换取经过审计验证的真实节水量与水资源回补效益（Water replenishment benefits），实现农户增收与企业可持续发展的双赢。 - 直接效益：根系靶向供给，大幅减少蒸发与渗漏损失。 - 综合增益：农作物显著增产，化肥、农药施用量减少，劳动力及能耗降低。 - 模式闭环：企业出资获取经核验的节水额度，农户获得技术升级与经济收益。 --- ### [08:55](https://www.youtube.com/watch?v=Po_UagChdlA&t=535s) 水资源管护项目的全流程实施与多维价值创造一个标准的水资源管护项目通常遵循严密的闭环流程：首先，企业选定自身已有业务或计划扩张的目标流域，根据自身抵消需求测算达到“水中和”（Water neutral）甚至“水正效益”（Water positive）所需的回补水量；随后，耐特菲姆负责组织并招募当地仍在使用传统漫灌方式的农户加入计划。项目推进的关键在于建立基于实测数据的严谨基准线（Baseline），拒绝模糊估算。耐特菲姆通过现场勘测记录历史与当前状态，部署精准灌溉优化系统及持续监测设备，对田间每年的实际节水量进行常态化监测与核验。这种机制为科技企业提供了极具公信力的水正效益进展证明，使企业在面临干旱气候或公众审视时具备高度的韧性，消除因缺水导致数据中心停运或无法扩建的后顾之忧。此外，精准灌溉还能显著减少农田温室气体排放，进一步放大生态与社区效益。 ```mermaid flowchart LR accTitle: 流域水资源管护项目落地实施全流程 subgraph 阶段一["目标确立"] A["企业选定目标流域"] --> B["测算水正效益回补指标"] end subgraph 阶段二["农户招募与基线构建"] B --> C["耐特菲姆招募漫灌农户"] C --> D["建立实测用水基准线 (Baseline)"] end subgraph 阶段三["技术部署与效益验证"] D --> E["部署精准灌溉与监测硬件"] E --> F["年度持续节水核验与韧性交付"] end ``` - 流域选定与核算：根据企业运营足迹设定水正效益补水量。 - 数据基准线：基于实际传感器与现场数据建立，杜绝单纯理论估算。 - 溢出效应：缓解共享流域用水压力、减少温室气体排放、提升农户抗旱韧性。 --- ### [11:51](https://www.youtube.com/watch?v=Po_UagChdlA&t=711s) AI 赋能精准灌溉：GrowSphere 软硬件一体化操作系统为了让精准灌溉发挥极致效能，AI 技术在田间管理中构筑了闭环控制。耐特菲姆推出了名为 GrowSphere 的农场操作系统，旨在通过自动化控制、日常全方位监测和智能决策建议，帮助农户实现水肥等生产要素的最优配置。 GrowSphere 系统整体由两层架构组成：底层是扎根于田间的硬件系统，涵盖滴灌管线、控制阀门、过滤系统、泵站等灌溉基础设施，以及监测土壤墒情、气象和水力管网压力、流量的各类传感器；上层则是将传感器与控制器互联的智能软件与 AI 模型，实时接收环境与水力数据，输出每日可执行的农事灌溉建议，实现从数据采集到智能执行的全流程闭环。 - 核心平台：耐特菲姆 GrowSphere 农业操作系统。 - 闭环能力：田间全自动控制、农艺与水力状态全天候监测、每日动态农事建议。 - 硬件基础设施：包含管道、阀门、泵站、过滤组件，以及农艺传感器和管网水力传感器（压力、流量）。 ### [13:23](https://www.youtube.com/watch?v=Po_UagChdlA&t=803s) 云端架构与超本地化 AI 作物模型演讲者详细拆解了系统的云端软件架构层。该软件层向上连接了海量的外部数据源，包括气象预报、卫星图像以及无人机航拍数据，向下则整合农田终端上传的各类数据。所有这些异构信息最终汇流进入核心的作物模型（Crop Models），模型根据农田里瞬息万变的真实动态，每日向农户输出具体的灌溉与施肥（Fertigation）操作建议。系统的核心 AI 能力主要体现在模型的持续进化与“超本地化”（Hyper-local）特性上。演讲者特别指出，许多行业外的人往往不理解农业用地的特殊性——每一块田地都是完全不同的。即使是紧挨着的两块相邻地块，由于历史种植作物不同、过往土壤管理与改良方式各异，其土壤结构与保水能力也可能存在天壤之别。因此，粗放通用的建议（比如“昨天下了雨，今天就别浇水”）在现代农业中价值极其有限；真正能带来效益的，正是这种基于超本地化洞察构建的 AI 作物模型，并且它能够在各个生长季之间持续迭代与演进。 ```mermaid flowchart TD accTitle: 精准灌溉云端 AI 系统架构流程图 subgraph InputLayer["数据输入层"] W["气象预报与外部环境"] S["卫星与无人机遥感影像"] F["田间传感器与液压设备"] end subgraph BrainLayer["云端 AI 决策核心"] M["超本地化 AI 作物模型<br/>（跨产季持续演进）"] A["LLM Agent 智能管家<br/>（自然语言答疑与自动化决策）"] end subgraph OutputLayer["执行与建议"] D["每日精准灌溉与施肥策略"] C["阀门与控制器自动调度"] end W --> M S --> M F --> M M --> A A --> D D --> C ``` - 作物模型并非静态分析公式，而是基于 AI 训练并跨季度持续学习迭代。 - “超本地化”洞察击破了传统农业管理盲区，针对相邻地块因历史耕作习惯不同所导致的微观差异实施针对性调配。 - 摆脱“下雨即停灌”等粗线条规则，实现对土壤含水量、气象走势和作物生长阶段的动态全量拟合。 --- ### [14:41](https://www.youtube.com/watch?v=Po_UagChdlA&t=881s) 农户的认知负荷与内置 LLM Agent 自动化方案演讲者坦言现代农业的运作极其复杂，农户面临着巨大的认知负荷。一个现代农户如果想管好农田，几乎必须成为全能专家：他们不仅要精准把握灌溉量，还得权衡种子的选用、植保农药的喷洒窗口期以及最佳收割时机。面对当下日益多变的极端气候与市场环境，每天要做的繁复决策常常令人心力交瘁。为了给农户彻底减负，Netafim 在系统中嵌入了基于大语言模型（LLM）的智能体（Agent）系统。该 LLM Agent 就像农户身边的数字农业助理，不仅能用自然语言清晰解释田间正在发生的各种异常情况，更能将复杂的决策流程转化为自动化执行指令。农户不再需要事必躬亲地奔波于各个地块之间逐一巡视排查，而是可以通过智能体实现高度自动化的农田运营。 - 现代农艺决策维度繁杂，涵盖灌溉控制、种子选育、病虫害植保、采收节点规划等多个专业领域。 - 系统内置 LLM Agent，帮助农户透视并理解各类传感器警报和数据波动的深层原因。 - 通过智能体驱动底层执行，将农户从高强度的手动巡田与繁琐操作中解放出来。 --- ### [15:25](https://www.youtube.com/watch?v=Po_UagChdlA&t=925s) 现场短片演示：GrowSphere OS 的田间协同实操为了更直观地展示技术在真实农场场景中的应用，现场播放了一段短片。短片生动还原了农场管理团队日常的焦虑与混乱：气温突然预测升高 3.4 摄氏度，21 号地块的作物看起来干渴打蔫，而 12、15、33 号地块的土壤含水量却偏高；一位农场工人甚至准备花 5 个小时开车前往偏远的北部地块巡视，团队内部还在为“当前到底是在过度灌溉还是灌溉不足”争论不休。 ```mermaid flowchart LR accTitle: GrowSphere 操作系统集成架构 subgraph Platform["GrowSphere OS 操作系统"] Core["中央控制与监测平台"] end subgraph Connectivity["硬件与感知互联"] P["灌溉施肥计划"] C["中央控制器"] H["液压系统"] S["田间传感器网络"] end subgraph Terminal["多端协同与交互"] Advisor["Crop Advisor 全球作物库"] UI["手机 / 平板 / 电脑随时接入"] end Core --- P Core --- C Core --- H Core --- S Core --> Advisor Core --> UI ``` 短片随后展示了破解之道——GrowSphere 操作系统（GrowSphere OS）。作为灌溉与水肥一体化管理的中枢，它将灌溉方案、田间控制器、液压管路系统以及土壤传感器无缝串联在一起： - **全地块远程实时感知**：无需耗费数小时驱车前往偏远地块，坐在办公室即可通过手机、平板或电脑屏幕查阅所有地块的墒情和灌溉状态，工人打趣再也不用“受罪跑去偏远的北部地块”。 - **闭环故障自愈与计划重排**：当 7 号地块由于计划灌溉未完成而触发传感器干旱警报时，系统不会被动报错，而是自动重写并更新了次日的灌溉日程以补偿水分缺口。 - **突发天气自适应补偿**：一旦遭遇突发暴雨，系统会第一时间把降雨补给量自动折算进后续的灌溉方案中，杜绝水资源浪费。 - **全球作物数据库支持（Crop Advisor）**：即使农场尝试引种从未接触过的全新作物品种，系统也能调用全球类似环境下的作物数据与模型给出最佳管护建议。 --- ### [17:45](https://www.youtube.com/watch?v=Po_UagChdlA&t=1065s) 结语：闭合 AI、水资源管理与农业的共享流域闭环演讲最后，Max Moldavsky 回到了本次演讲的核心主旨。他明确指出，水资源如今已经成为整个人工智能产业与数字化基础设施不可回避、必须参与讨论的核心议题。这不仅关乎数据中心的冷却与扩张，更关乎数字世界与物理实体世界的利益平衡。演讲者总结道，AI 带来了我们所需的超强智能与增长引擎，而精准灌溉技术则将这种智能转化为了守护水资源（Water Stewardship）的切实行动。这种立足于行动的水资源管理，最终在“AI、粮食保障、现代农业”之间闭合了生态闭环。他呼吁行业各方必须站在同一流域的视角，通过智能优化的调配方案管理共享水资源，共同铸就数字时代与农业生产共生共荣的未来。全场在热烈的掌声中落下帷幕。 > “AI 带来了智能与增长，精准灌溉提供了支持水资源管理的实际行动，而水资源管理正是闭合 AI、粮食、农业与共享未来之间闭环的关键纽带——这一切都依托于在同一本地流域内，以最优化的方式为所有人管理好水资源。” ## Notable quotes > [00:04](https://www.youtube.com/watch?v=Po_UagChdlA&t=4s) 推动这一议题的核心绝非单纯的公关宣传（PR）、市场营销或品牌口号，而是关乎企业的实际“韧性”（Resilience）——使企业在业务连续性不受威胁的前提下持续运营，赢得周边社区的信任，并获得至关重要的“业务扩张许可”（License to grow）。 — Max Moldavsky > [01:47](https://www.youtube.com/watch?v=Po_UagChdlA&t=107s) 预计到 2050 年，全球粮食需求将比现在增加 60% 至 70%。这使得数据中心无法再将自身视为孤立的数字实体，其建设与运营必然会触碰当地资源的敏感神经。 — Max Moldavsky > [01:47](https://www.youtube.com/watch?v=Po_UagChdlA&t=107s) 科技巨头而言，确保在不发生断水风险的前提下持续扩容，已成为核心竞争力之一，他们必须学会在所在流域扮演“好邻居”（Good neighbor）的角色。 — Max Moldavsky > [04:35](https://www.youtube.com/watch?v=Po_UagChdlA&t=275s) 仅仅提高数据中心自身的冷却水利用效率是远远不够的，这类内部工程优化存在物理极限。企业必须将视野扩展到“水资源管理”（Water Stewardship），通过出资支持能够在当地流域产生可量化、可验证节水效果的外部项目，从而弥合水资源缺口。 — Max Moldavsky > [07:26](https://www.youtube.com/watch?v=Po_UagChdlA&t=446s) 耐特菲姆将其核心价值主张概括为“以更少资源产出更多”（Grow more with less）。 — Max Moldavsky > [08:55](https://www.youtube.com/watch?v=Po_UagChdlA&t=535s) 项目推进的关键在于建立基于实测数据的严谨基准线（Baseline），拒绝模糊估算。 — Max Moldavsky > [13:23](https://www.youtube.com/watch?v=Po_UagChdlA&t=803s) 每一块田地都是完全不同的。即使是紧挨着的两块相邻地块，由于历史种植作物不同、过往土壤管理与改良方式各异，其土壤结构与保水能力也可能存在天壤之别。 — Max Moldavsky > [13:23](https://www.youtube.com/watch?v=Po_UagChdlA&t=803s) 粗放通用的建议（比如“昨天下了雨，今天就别浇水”）在现代农业中价值极其有限；真正能带来效益的，正是这种基于超本地化洞察构建的 AI 作物模型，并且它能够在各个生长季之间持续迭代与演进。 — Max Moldavsky > [17:45](https://www.youtube.com/watch?v=Po_UagChdlA&t=1065s) AI 带来了我们所需的超强智能与增长引擎，而精准灌溉技术则将这种智能转化为了守护水资源（Water Stewardship）的切实行动。 — Max Moldavsky > [17:45](https://www.youtube.com/watch?v=Po_UagChdlA&t=1065s) AI 带来了智能与增长，精准灌溉提供了支持水资源管理的实际行动，而水资源管理正是闭合 AI、粮食、农业与共享未来之间闭环的关键纽带——这一切都依托于在同一本地流域内，以最优化的方式为所有人管理好水资源。 — Max Moldavsky ## Mentioned Max Moldavsky, Netafim, 耐特菲姆, GrowSphere, GrowSphere OS, Crop Advisor, Ai4, LLM Agent ## Video description Max Moldavsky - Director of Innovation and Climate Solutions, Netafim Water is becoming the defining constraint for both food production and digital infrastructure—so how do we manage it smarter, together? In this talk, I will explain how precision irrigation, powered by AI, turns water stewardship into measurable action at the basin level. By combining sensors, forecasts, and crop models, we can optimize every drop—reducing withdrawals, improving yields, and generating verifiable water savings. I will show how these auditable outcomes support corporate targets while strengthening shared-basin resilience. Join this session to explore how data-driven precision irrigation can align agricultural productivity with sustainable digital growth. [Source](https://www.youtube.com/watch?v=Po_UagChdlA)

Source quotes

[0:04] Max Moldavsky: 推动这一议题的核心绝非单纯的公关宣传（PR）、市场营销或品牌口号，而是关乎企业的实际“韧性”（Resilience）——使企业在业务连续性不受威胁的前提下持续运营，赢得周边社区的信任，并获得至关重要的“业务扩张许可”（License to grow）。 [1:47] Max Moldavsky: 预计到 2050 年，全球粮食需求将比现在增加 60% 至 70%。这使得数据中心无法再将自身视为孤立的数字实体，其建设与运营必然会触碰当地资源的敏感神经。 [1:47] Max Moldavsky: 科技巨头而言，确保在不发生断水风险的前提下持续扩容，已成为核心竞争力之一，他们必须学会在所在流域扮演“好邻居”（Good neighbor）的角色。 [4:35] Max Moldavsky: 仅仅提高数据中心自身的冷却水利用效率是远远不够的，这类内部工程优化存在物理极限。企业必须将视野扩展到“水资源管理”（Water Stewardship），通过出资支持能够在当地流域产生可量化、可验证节水效果的外部项目，从而弥合水资源缺口。 [7:26] Max Moldavsky: 耐特菲姆将其核心价值主张概括为“以更少资源产出更多”（Grow more with less）。 [8:55] Max Moldavsky: 项目推进的关键在于建立基于实测数据的严谨基准线（Baseline），拒绝模糊估算。 [13:23] Max Moldavsky: 每一块田地都是完全不同的。即使是紧挨着的两块相邻地块，由于历史种植作物不同、过往土壤管理与改良方式各异，其土壤结构与保水能力也可能存在天壤之别。 [13:23] Max Moldavsky: 粗放通用的建议（比如“昨天下了雨，今天就别浇水”）在现代农业中价值极其有限；真正能带来效益的，正是这种基于超本地化洞察构建的 AI 作物模型，并且它能够在各个生长季之间持续迭代与演进。 [17:45] Max Moldavsky: AI 带来了我们所需的超强智能与增长引擎，而精准灌溉技术则将这种智能转化为了守护水资源（Water Stewardship）的切实行动。 [17:45] Max Moldavsky: AI 带来了智能与增长，精准灌溉提供了支持水资源管理的实际行动，而水资源管理正是闭合 AI、粮食、农业与共享未来之间闭环的关键纽带——这一切都依托于在同一本地流域内，以最优化的方式为所有人管理好水资源。

Source quotes

Max Moldavsky: Water is becoming the defining constraint for both food production and digital infrastructure—so how do we manage it smarter, together? [0:04] Max Moldavsky: Shared water and shared future [0:04] Max Moldavsky: license to grow the business [1:47] Max Moldavsky: Will there be sufficient water in the local basin to sustain the community? [1:47] Max Moldavsky: Can rapid AI expansion coexist with the agricultural expansion required to feed the population? [1:47] Max Moldavsky: How can technology operators demonstrate authentic, responsible resource stewardship? [1:47] Max Moldavsky: good neighbor [7:26] Max Moldavsky: grow more with less [13:23] Max Moldavsky: it rained yesterday, so do not irrigate today [17:45] Max Moldavsky: Water is now an essential part of the conversation of AI that we need to have.

Original transcript

[0:04] So, today I'm going to speak with you [0:06] about an increased connection that is not always trivial between AI, food, and agriculture, [0:13] and it mostly comes around around water. And this is why you can see that is written here, [0:18] "Shared water and shared future." Because they are interconnected for future growth. [0:25] And one of the core messages that I want you to understand really from the beginning that [0:31] this entire story is not about PR. It's not about marketing, and not about just [0:37] messaging. It's about resilience. It's about the resilience to allow organizations to keep working [0:44] with no threats to their business continuity, with the community trust of everyone around them, [0:51] and not less importantly with the license to grow the business. [0:57] Now, AI for sure for everyone in this conference, we all understand that AI is the next great [1:03] revolution. And don't get me wrong, I think that the AI revolution is definitely a great one, [1:09] good one, but it also has an infrastructure layer. And infrastructure layers has consequences. [1:17] So, the physical footprint that AI have starts with compute, and the demand for compute is really [1:24] uh increasing exponentially as we as the years go and as we grow. [1:29] And this means that we need more power, more energy. And that mean also means that we need [1:35] more water for cooling the data centers, for the operation of the data centers. [1:40] And this is how water is starting to connect into the AI world. [1:47] Now, AI and food may seem like very different worlds, [1:53] and agriculture and AI may seem like very different worlds, but if you look at it [1:58] in a different perspective, you understand that both of these things are first of all essential [2:04] in different ways, but they also depend on the same local resources to succeed. [2:11] And the main resource that we're speaking about here and right now in the talk is about water. [2:17] Water is really at the core of everything. Water is the core of AI for operation, and [2:21] water is for sure at the core of agriculture and food that allows us to survive and to flourish. [2:29] But water is also an a finite resource and a diminishing one at that. Water is becoming [2:35] less and less available for any type type of use, whether it's again agriculture, AI, [2:41] or our own consumption as people. And that means that we need to optimize water usage for everyone. [2:51] But the problem doesn't stop only here. Uh AI is growing, but also the need for more food. [2:57] Population is growing. We need more We have more mouths to feed, and by 2050, [3:03] we are expected to need more than 60 or more 60 or 70% more food than what we have today in the [3:12] in the fact that we have diminishing resources to support that all. [3:18] That means that AI infrastructure cannot continue only to be some sort of a digital story. It [3:24] becomes a story of local resources, and the local here is the keyword. Every place that we have [3:33] new AI infrastructure, local stakeholders, whether they are farmers, whether they are residents in [3:40] the area, or any other local stakeholders, are beginning to ask the following question. [3:46] Will the water that we have in our local basin be enough for the communities. [3:55] Can this growth of AI coexist with growth agriculture that needs to happen to feed everyone? [4:02] And how the tech companies can be responsible and act more responsibly? This connects, of course, [4:08] with the main question the tech companies need to ask themselves, will we have enough water [4:12] to operate our businesses without any risk of discontinuation and with the risk of not [4:19] able to grow our compute? All of that makes the fact that one of the next competitive advantages [4:26] for tech companies is really to be seen as a good neighbor in the local basin where they operate. [4:35] And you may ask, okay, so let's just make the data centers very efficient in water and very efficient [4:42] in how they use water to cool down. That's true and it has to happen and it does happen, [4:48] but it's not enough. We really need to go to one step further, which is water stewardship. Go from [4:55] just using less water in the operation that can always have some sort of a finite limit to [5:01] how much impact it can bring to really improve the outcomes of everyone that is sharing the basin. [5:09] Uh and water stewardship is the way to bridge the gap and to bring verified outcomes by [5:17] funding projects that are actually saving significant amounts of water in the local basin. [5:26] There are different methods to save water and different methods to have an impact through water [5:31] stewardship. It can be from harvesting rainwater or investing in recycling graywater or even just [5:39] investing in upgrading water distribution systems and making them more efficient. But [5:47] probably the place with the highest potential for improvement and one of the biggest impact [5:52] is around agriculture and specifically around precision irrigation that can save [5:58] up to or sometimes even more than 50% compared to the ways that currently most of the farmers are [6:04] using water in their fields. And why agriculture? Why should we focus on that for water stewardship? [6:13] And the answer is pretty simple. With all the growth that we see with the AI, [6:18] with all the growth that we see in other industries, agriculture is still responsible [6:22] for around 70% of the consumption of fresh water globally. That means that [6:29] here's where the potential lies. Here is where we can bring the most impact. And bad news, [6:36] but also good news in terms of potential, is that there is a lot of room for improvement. [6:42] Even today, around the world, around 75% of the irrigated land are irrigated in what we call [6:49] flood irrigation. This is the same method that is used around the world for centuries and centuries. [6:56] And like you can understand, it's probably not a very efficient way. You just flood your field and [7:01] the water goes to the plant, but also just wastes and go to the soil and not where we want it to go. [7:08] Precision irrigation is still only 7% of the irrigated lands. That means that we have [7:13] so much potential to save on water and improve efficiency and productivity through increasing the [7:20] penetration of precision irrigation. For for the ones of you that don't really know what precision [7:26] irrigation means, so it really means to bring the water directly into the root zone of the plant, [7:33] not just there somewhere around on the soil. And the fact that it goes directly into the root zone [7:42] brings the many very inherent advantages. So, water savings are obvious cuz that's what we're [7:48] discussing, but alongside that, a very significant increase in yields and output of the growers, [7:56] reduced costs of inputs like fertilizers, crop protection, labor, energy, and others. [8:02] And all of this, of course, combined into improved sustainability. We like to call [8:07] it grow more with less. And this less part is super crucial when we think about water. So, [8:14] the value proposition that precision irrigation brings to water stewardship is the ability [8:20] to provide with the best return on investment, huge amounts of verified water reduction and [8:28] water replenishment benefits to businesses that want to invest in their local basin [8:34] through funding technologies to growers that want to advance how they do their irrigation today. [8:43] So, the businesses get the benefit, the farmers get are getting the technology that helps [8:48] them to succeed. Uh and all of this through the precision irrigation technologies and solutions. [8:54] How such a program looks like in actual uh in actuality? So, a business selects the basin where [9:02] it wants to invest. It can be a place where they already have a footprint or a place where they [9:08] want to increase their footprint. This is analyzed into amount of water that is actually needs to [9:14] be replenished to keep the business water neutral or even water water positive. After that, Netafim [9:24] is taking care of enrolling farmers that are currently using predated solutions to join the [9:31] program and to get the funding to basically improve and to get uh precision irrigation [9:37] technologies in their field. Once these farmers are enrolled, we are establishing the baseline [9:44] of what is actually happening today, right? You have to put a lot of focus and emphasis on [9:49] actual numbers and not just estimations. So, we're doing a baseline of the current state of affairs, [9:57] implementing the technology needed for both the optimization of the irrigation, but also [10:02] for the ongoing monitoring that is needed more for the grower, but also for the companies that [10:08] want to see the actual water reduction year over year. And that is what brings the benefit. Now, [10:18] it all entails into fact that the companies getting, first of all, credible water positive [10:24] progress. They have goals, they have targets, and this is what they're getting in a very verified [10:31] and a strong way. But, and let's take you for a second to the beginning of my talk, [10:36] it's all about resilience. They're getting much more resilience to droughts in the area, [10:43] to any scrutiny that can come. They were able to operate in the basin without the concern of [10:50] I may not have enough water for my data center to operate correctly or to increase my capacity in [10:57] the same location. It brings the community trust. It's a positive story that helps the farmers, [11:03] the local stakeholders, and everyone in the basin to understand that we are here together [11:08] in our shared future with water. And all in all, it takes us to basically giving the [11:16] uh the data center companies, the tech companies, their license to grow more and more. [11:23] The benefits go beyond water. Um it touches a lot upon the positive environmental impact because [11:31] the inherent value of precision irrigation that reduces significantly greenhouse gas emissions, [11:38] which are also an important part of what we want to achieve, farmer resiliency, [11:44] and of course the community benefit of less pressure on uh the shared local sources. [11:51] Now, we spoke a lot about water, but let's pick also on how AI basically touches [11:58] and closes the loop in precision irrigation and helps growers to achieve better results. So, [12:06] in our case, it's through our own system that is called the GrowSphere. This is basically an [12:10] operating system for growers to optimize their water use and their other input uses [12:18] through a closed loop of control and automation of everything that they're able to do in the field, [12:24] ongoing daily monitoring of everything that is happening from an agronomical and hydraulic [12:30] uh perspective, and maybe the most importantly, recommendations to give them actual [12:38] uh action items on a day-to-day basis to make them succeed uh with their operation. [12:45] How does the system looks like? So, basically, it's a two-layer uh system. First of all, [12:52] it's the hardware on the ground, starting from the irrigation system itself, from the pipes [12:58] to the valves, filters, pumps, anything needed for precision irrigation to succeed, but also [13:05] uh monitoring equipment like sensors, whether they're agronomical sensors, whether they're [13:10] hydraulic sensors that are checking what is happening with pressures uh and flow, and so on. [13:17] All of this is connected to the controllers that control everything that is happening here, [13:23] and to the cloud layer, the software layer, where it is connected to external sources like weather, [13:29] like images from satellites or drones. And of course, [13:34] the crop models that are taking all of this input and are giving daily recommendation based on what [13:41] is actually happening to the farmer of what you need to do with your irrigation fertilization. [13:48] Uh in terms of what key AI elements we have in the system, so of course first of all the crop models [13:55] are not just analytical, they're based they're AI based and they are constantly evolving to [14:02] uh better perform from a season to a season. But not less importantly, they are hyper local. [14:09] Something that people not always understand about agriculture and land in general [14:15] is that every plot is different. A plot of someone can be significantly different from [14:22] the land of his neighbor and can be from variety of historical reasons, from what was grown there, [14:27] how it was treated and so on. So this hyper local insights are what actually gives the benefit [14:34] beyond just generic recommendation of okay, yesterday it rained, so don't irrigate today. [14:41] Uh and the last thing is that agriculture is super complex. A farmer needs to be an expert on [14:50] how to irrigate, which seeds to use, when to use crop protection, when to harvest, [14:56] and many, many other decisions that are just super complex in today's world. And what we are doing in [15:03] our system is to provide them built-in LLM agent-based system that helps them [15:11] to understand what actually going on in their system, in their field, uh and to provide them [15:18] the ability to automate these activities and not the need to do everything by themselves. [15:25] Um so, this is how the system looks like, but a short video can help understand even better. [15:34] Hey everyone. So, how are we doing with our irrigation? Uh plans are almost ready. Just [15:42] need a quick update cuz it's going to be hotter than we thought. 3.4° higher, to be precise. Yeah, [15:49] some of the crops in plot 21 are looking a bit thirsty. And soil water content in plots 12, 15, [15:57] and 33 is too high. I'm heading to the northern section. I'll report back in about 5 hours. [16:02] This is a mess. Hey, are we over or are we under irrigating right now? Under. Definitely over. [16:08] Why does it have to be so complicated? Actually, it doesn't. We can use GrowSphere OS. [16:15] What's that? GrowSphere is an operating system for irrigation and fertigation that helps us to plan, [16:20] control, and monitor everything all in one place. Really? How? GrowSphere connects [16:27] our irrigation plans to the controller, to the hydraulic system, to the field sensors, [16:31] giving us real-time irrigation recommendations. Okay. Let's see what [music] it can do. [16:37] All right, check it out. We can see the irrigation status of all our plots at [music] any time. [16:43] Without having to drive out to the field? Yep. Awesome, cuz I hate driving to the northern [16:49] section. Hold on. I see that there's an alert on plot seven, incomplete scheduled irrigation. [16:54] And I see the sensors are showing dry soil. No worries. GrowSphere has already updated [16:59] tomorrow's irrigation schedule accordingly. But what if there's a sudden rainstorm? [17:03] GrowSphere automatically factors this into [music] our irrigation plans, too. Hm, [17:08] that's pretty smart. What if we start growing a variety we haven't grown before? [17:12] GrowSphere's crop advisor can help us with that, too, using data from similar crops grown around [17:17] the world. You can do all of this from anywhere? Anywhere. Your computer, tablet, even your phone. [17:24] Fantastic, cuz You hate the northern section. So, GrowSphere, huh? GrowSphere. GrowSphere! [17:37] I think our crops are going to love it. [17:44] [music] [17:45] So, just to close up the talk, my closing message is really simple. [17:52] Water is now an essential part of the conversation of AI that we need to have. [17:59] AI really provides the intelligence, provide the growth that we need. Precision irrigation provides [18:07] the action to support the water stewardship, and the water stewardship is what provides [18:12] the connection that closes the loop between AI, food, agriculture, and the shared future [18:18] that all relies upon an optimized way of managing water for everyone in the same local basin. [18:26] Um that's it. Thank you, everyone. Hopefully you enjoyed. [applause]