%0 Report %A Naudé, Wim %A Sequeira, Tiago Neves %A Almeida, Derick %T From Automation to Acceleration: Transitional Regimes in AI-Driven Growth %D 2026 %8 2026 Nov %I Institute of Labor Economics (IZA) %C Bonn %7 IZA Discussion Paper %N 19110 %U https://www.iza.org/publications/dp19110 %X Can AI turn automation into a sustained acceleration of economic growth? This paper argues that the answer depends on transitional dynamics. It develops a semi-endogenous growth model that separates what AI can do in principle, from how much research output actually responds to AI , and both from whether growth actually accelerates due to transformative AI. It shows that the date at which AI starts to matter for research is the outcome of a race: growing knowledge widens the set of research tasks that AI can perform, which loosens the human bottleneck, while faster accumulation of AI capital adds AI input faster than tasks open up, which tightens it. Calibrated to U.S. task-automation shares and hardware energy-efficiency trends, four scenarios show that the same economy can experience no acceleration, a modest transitory acceleration, or an explosive surge that is ultimately curbed by energy limits. Under strong task complementarity, early acceleration is driven by task reallocation rather than by AI-capital accumulation. Near-complete automation is neither necessary nor sufficient for permanent explosive growth; the outcome depends on the substitutability of AI and human researchers, and on how hard ideas are to find. %K artificial intelligence %K automation %K economic growth %K ideas production function %K transformative AI %K transitional dynamics