Against the background of severe farmland fragmentation and continuous rural labor outflow in developing countries, land consolidation through land transfer has long been regarded as the primary path to realize large-scale agricultural production. However, land transfer faces institutional and market constraints in smallholder-dominated regions. This paper takes China’s full-process agricultural production service pilot policy as a quasi-natural experiment, adopting 2006–2023 provincial panel data and staggered DID to explore how service-oriented scale reshapes cultivated land use patterns and boosts staple grain supply. We find that agricultural production services raise total grain output mainly via expanding cultivated land sown area (extensive margin) rather than lifting per-unit land yield. By popularizing mechanized operations in labor-intensive sowing and harvesting stages, the policy alleviates seasonal labor constraints, raises per capita cultivated acreage, increases multiple cropping index and optimizes grain-oriented cropping structure, thus mitigating farmland abandonment. Although the policy significantly promotes rural land transfer, the pre-existing land transfer level cannot strengthen its grain-promoting effect, which verifies that service-based scale can operate without large-scale land right consolidation. Further evidence shows the policy also improves agricultural total factor productivity, realizing coordinated optimization of land, labor and machinery factor allocation. This study supplements the land economics literature by distinguishing two scale development paths (land transfer vs service aggregation), and provides a land governance reference for smallholder economies to stabilize grain security under fragmented farmland.