塑造农业的深科技革命_42页_20mb

Summary of "Shaping the Deep-Tech Revolution in Agriculture"

Core Content

This report, published by the World Economic Forum in November 2025, explores the transformative role of deep technology (deep tech) in future-proofing global agriculture. It identifies key challenges facing the sector and highlights seven promising deep-tech domains that can help address them. The report emphasizes the need for ecosystem building and cross-sector collaboration to scale these innovations effectively.

Main Challenges in Global Agriculture

• Shrinking workforce: Agricultural labor is declining due to urban migration and aging populations, with the global average farmer age around 60 years.
• Climate change and weather extremes: Projected to reduce crop yields by 24% by 2100 if emissions are not curtailed.
• Natural resource degradation: Agriculture accounts for 70% of freshwater withdrawals and is heavily dependent on soil, which is expected to be 90% degraded by 2050.
• Rising food demand: Global food production needs to increase by about 70% by 2050, with a near doubling required in developing countries.
• Demand-supply mismatch: Leads to significant food loss and waste, with 13% lost at the farm level and 19% wasted in retail and households.
• Geopolitical instability: Threatens food supply chains, as seen in the 2022 fertilizer price surge due to the Russia-Ukraine conflict.

Key Deep-Tech Domains in Agriculture

The report identifies seven key deep-tech domains with significant potential for agricultural transformation:

1. Generative AI (GenAI)

• Use cases: Customized farmer advisory, pest management, market price prediction, agentic AI systems, and climate risk simulations.
• Drivers: Advances in large language models (LLMs) and the availability of agricultural data.
• Barriers: Data quality issues, hallucinations and errors, and limited transferability across regions and domains.
• Case Study: The Wadhwani Institute for Artificial Intelligence developed AgriAI Collect, using speech recognition and LLMs to improve data collection in smallholder farming.

2. Computer Vision

• Use cases: Pest and disease identification, crop stress monitoring, autonomous grading and sorting, and navigation for agricultural robots.
• Drivers: Lower sensor costs, improved access to mobile cameras, and advances in deep learning models.
• Barriers: On-field variability, lack of training data, and high investment costs.

3. Edge Internet of Things (Edge IoT)

• Use cases: Real-time on-farm data processing, autonomous decision-making for irrigation and fertilization, and farm automation.
• Drivers: Need for low-latency and low-bandwidth data processing in areas with poor connectivity.
• Barriers: High capital costs for farmers and limited interoperability among edge systems.

4. Satellite-enabled Remote Sensing

• Use cases: Continuous, large-scale monitoring of farm conditions, aiding in data-driven decisions.
• Drivers: Enhanced spatial and spectral capabilities, and increased data capture frequency.
• Barriers: Limited accuracy in small and fragmented farmlands, and in multi-cropping scenarios.

5. Robotics (including Drones)

• Use cases: Automation of labor-intensive tasks such as precision planting, weeding, and harvesting.
• Drivers: Advances in AI-enabled perception and cloud-edge integration.
• Barriers: High capital costs, limiting adoption in low-wage, labor-abundant countries.

6. CRISPR

• Use cases: Development of crops with enhanced traits such as drought tolerance and pest resistance, bypassing traditional breeding cycles.
• Drivers: Precision and speed of genetic editing.
• Barriers: Regulatory hurdles and public perception issues.

7. Nanotechnology

• Use cases: Precision nutrient and pesticide delivery, reducing input use and environmental impact.
• Drivers: Wide range of applications from nutrient management to biosensing.
• Barriers: Lack of research on long-term environmental and health impacts, hindering scalability.

Breakthrough Use Cases

• Swarm robotics: Coordination of multiple robots for efficient farm operations.
• Agentic AI systems: Autonomous decision-making systems that adapt to farm conditions.
• Macro crop planning: AI-driven strategies for large-scale agricultural planning.
• Supply chain risk-assessment tools: Predictive models to manage risks in food supply chains.

Optimizing Agri Deep-Tech Ecosystems

• Collaboration: Essential across governments, research institutions, start-ups, corporations, and financiers.
• Policy and finance: Supportive policies and patient capital are crucial for ecosystem development.
• Digital infrastructure: Robust data systems and internet connectivity are needed for widespread adoption.
• Human capital: Skilled labor and training programs are required to implement and maintain deep-tech solutions.

Conclusion

The report concludes that deep tech is vital for future-proofing agriculture, offering transformative potential in addressing climate change, resource scarcity, and food security. However, scaling these technologies requires a collaborative, ecosystem-based approach that includes innovation support, policy development, and investment in digital infrastructure. It serves as a call to action for stakeholders to work together in driving the deep-tech revolution in agriculture.

Annexe: Methodology for Selecting Technology Domains

The report outlines the methodology used to identify the seven key deep-tech domains, which includes community consultations, agritech convenings, and on-the-ground pilots. These insights are enriched by structured interviews with deep-tech experts.

Contributors

The report is authored by the World Economic Forum in collaboration with Microsoft's Agri-Food division and other stakeholders. It reflects the collective insights of the AI4AI initiative, which has been working since 2021 to promote agritech innovation and adoption.