Key message Dynamic global vegetation models are key tools for interpreting and forecasting the responses of terrestrial ecosystems to climatic variation and other drivers. They estimate plant growth as the outcome of the supply of carbon through photosynthesis. However, growth is itself under direct control, and not simply controlled by the amount of available carbon. Therefore predictions by current photosynthesis-driven models of large increases in future vegetation biomass due to increasing concentrations of atmospheric CO₂ may be significant over-estimations. We describe how current understanding of wood formation can be used to reformulate global vegetation models, with potentially major implications for their behaviour.