Policy sequencing and the capability trap: AI governance instrument design across Asian developing countries

Does the order in which developing Asian countries adopt AI policy instruments shape governance capacity outcomes? Existing AI governance scholarship, anchored in advanced-economy experience, assumes that regulatory frameworks should precede or accompany other policy instruments. This article tests that assumption against evidence from 29 developing Asian countries observed annually from 2020 to 2025 through the Government AI Readiness Index. Sequence classification, OLS regression, and fuzzy-set Qualitative Comparative Analysis assign countries to four sequence types according to whether governance-pillar or capacity-pillar improvements came earlier. A Governance – Capacity Gap metric tracks the distance between institutional form and operational capacity. Countries that adopted regulation ahead of capacity investment accumulated widening gaps; those directing resources towards infrastructure and human capital before formalizing regulation converged towards balanced profiles and stronger composite scores, robust to a capacity-only dependent variable. The fsQCA identifies two sufficient pathways: both require pre-existing infrastructure, one combining it with non-upper-middle income (under which regulation-first sequencing is also compatible) and the other with investment-oriented sequencing in South and Central Asia. Infrastructure preconditions, more than sequencing, determine whether capacity improvement is realized. The findings carry implications for international organizations framing AI governance prescriptions and the policy design literature on instrument-mix sequencing.