In water-limited ecosystems, evidence suggests that photovoltaic (PV) arrays may alleviate water stress, potentially mitigating the loss of key ecosystem services such as forage production. This hypothesized aridity mitigation potential (AMP) of PV arrays posits that negative effects of reduced sunlight on photosynthesis and aboveground net primary production (ANPP) are offset by improved water relations, maintaining or enhancing productivity in arid and semi-arid ecosystems. Field tests of this hypothesis are lacking, however. Here we test the AMP hypothesis in a semi-arid Colorado (USA) grassland by contrasting ANPP responses to a single-axis tracking PV array in years with above- and below average precipitation. At all sites within and outside the PV array, precipitation inputs strongly limited ANPP. But in dry years, ANPP was increased by ∼20% in the array relative to adjacent open grassland, with ANPP in some array locations (near PV panel edges) increased by ∼90%. This AMP effect was muted with average precipitation and disappeared in wet years. We conclude that with climate change projections of increased aridity and drought, PV arrays that reduce plant water stress could provide a unique opportunity to reduce greenhouse gas emissions and mitigate some of their ecosystem consequences in water-limited grasslands.