Smart factories often combine advanced automation, connected equipment and recognised smart-manufacturing status, yet their invention performance remains uneven. This study examines why additional automation infrastructure does not always raise invention output.
We analyse 85 listed-firm hosts in China's national smart-manufacturing demonstration programme. Fiscal-year 2022 technology, organisational and environmental conditions are linked to 2024 invention patent intensity. Necessary condition analysis identifies ceiling constraints on attainable invention, while fuzzy-set qualitative comparative analysis identifies configurations associated with high and near-zero invention.
Infrastructure intensity does not form a statistically supported ceiling on attainable invention. Instead, human capital and the organisational resource base constrain the upper frontier of invention. High invention can emerge through multiple capability-centred configurations, including pathways that do not require further infrastructure deepening. Near-zero invention is associated with asymmetric configurations, especially infrastructure-heavy profiles lacking sufficient human-capital support.
The study contributes to manufacturing technology management by shifting the explanation from automation adoption to automation-to-invention conversion. It theorises capability ceilings and infrastructure–capability mismatch as mechanisms explaining why advanced smart factories can show near-zero invention performance, and it offers a stage-gated sequencing rule for automation investment.
