Agriculture is increasingly challenged by human‐induced climate change, land degradation and increasing crop losses caused by pests and pathogens. Reductions in crop yields of 17–30% across major crops unevenly affect food insecure regions. Yet while digital tools from basic farm-management apps to sophisticated automation have been spreading, cost, connectivity and expertise remain significant barriers preventing such technology from becoming more widespread. Incorporating such technologies into full digital twins has the potential to link disparate data streams, improve decision-making reliability and deliver interventions that are specifically relevant to a locale. Yet most agricultural digital twins are still confined to experimental scale due to the difficulty of modelling life, infrastructure deficits in rural areas and difficulties in ensuring robustness of the model under environmental diversity. These gaps will be filled with the development of standardized protocols for data, open access archives and repositories, large-scale interdisciplinary collaboration, and focused training efforts. As governments and innovators seek to reconcile yield with sustainability, digital twins might offer a way to truly precise, adaptive and resilient farm management provided the technical, economic and policy challenges are addressed in concert.