Executive summary The agricultural sector globally faces converging pressures: a shrinking workforce, intensifying climate extremes, natural resource degradation, rising food demand and geopolitical instability. These challenges threaten food security and rural livelihoods, demanding transformative action. Novel science-backed technologies often referred to as deep tech could drive much of this action in the coming decades. This report explores deep tech’s potential in agriculture and identifies seven promising deep-tech domains as pivotal for tackling current and future agricultural challenges. These are: Generative AI (GenAI): Offers use-cases ranging from tailored farmer advisory and pest management to agentic AI systems and climate risk simulations. GenAI’s applicability in agriculture is driven by recent advances in large language models (LLMs) and the increasing availability of agricultural data. However, despite these advances and the growing adoption of GenAI, the lack of high-quality data for training hyperlocal models remains a barrier to usability. Computer vision: Provides use cases such as rapid pest and disease identification or plant stress detection. The growth of computer vision use cases has been fuelled by decreasing camera costs and advances in deep-learning models. However, unlike in industrial units, on-field variability (for instance, variations in on-field lighting and plant appearance between growth stages) restricts its applicability for agriculture. Edge internet of things (IoT): Enables real-time, on-farm data processing and autonomous decision-making for agriculture. This minimizes latency and bandwidth dependency, especially in areas with poor internet connectivity. Edge IoT can improve decisions related to irrigation, fertilization and disease management, while automating farm processes. The domain currently faces challenges, with high capital costs for farmers and limited interoperability among edge systems. Satellite-enabled remote sensing: Allows continuous and large-scale monitoring of farm conditions at affordable costs, aiding data-driven decision- making. Enhanced spatial and spectral capabilities and increased data capture frequency are driving adoption in agriculture, although the level of accuracy is limited in small and fragmented farmlands or when multi-cropping is practised. Robotics: Permits the automation of labour-intensive tasks such as precision planting, weeding and harvesting. Advances in AI-enabled perception and cloud-edge integration are driving its adoption. However, high capital costs limit its uptake in low-wage, labour- abundant countries. CRISPR: Accelerates the development of crops with enhanced traits such as drought tolerance and pest resistance, bypassing lengthy traditional breeding cycles. The potential precision and speed of CRISPR-based editing are significant drivers of use, but regulatory approval processes and negative public perception are barriers to commercialization. Nanotechnology: Offers precision in nutrient and pesticide delivery, reducing input use and environmental impact. It enables a wide range of use cases, ranging from pest and nutrient management to controlled release of inputs to biosensing, though a lack of research data on the long-term environmental and health impacts remains a barrier to scale. This report identifies breakthrough agri deep-tech use cases derived from these domains. As several are yet to be commercialized, it further provides recommendations to optimize support for agri deep tech. It elaborates collaborative efforts in policy, finance, human capital, data/digital infrastructure and innovation support to seed promising agri deep-tech ideas, de-risk innovations and enable impact at scale.Deep tech has the potential to future-proof agricultural systems, but collaboration is critical to deliver them at scale. Shaping the Deep-Tech Revolution in Agriculture 4

Shaping the Deep Tech Revolution in Agriculture 2025