U.S. Productivity Growth Doubles as AI Enters Economic Harvest Phase
Palo Alto, Monday, 16 February 2026.
Analysis reveals 2025 productivity growth hit 2.7%, nearly doubling the prior decade’s average. This surge signals the economy has officially transitioned from AI experimentation to structural utility.
The J-Curve and the Harvest Phase
According to Erik Brynjolfsson, Director of Stanford’s Digital Economy Lab, the U.S. economy is emerging from the investment-heavy trough of the “J-curve”—a historical pattern where productivity gains lag behind initial capital investments before surging. Brynjolfsson argues that the data from 2025 indicates the economy has now entered a “harvest phase,” where the massive expenditures in generative AI infrastructure and training began to manifest as measurable economic output [1][2]. This transition suggests that the integration of artificial intelligence has moved beyond pilot programs, allowing businesses to reap the rewards of reorganized workflows and workforce retraining [2].
Quantifying the Productivity Surge
The statistical evidence for this shift is stark. Brynjolfsson’s analysis highlights that U.S. productivity rose to approximately 2.7% in 2025, a figure that stands in sharp contrast to the 1.4% annual average recorded over the past decade [1][2]. This represents a productivity rate that is 1.929 times higher than the recent historical norm. Stephen Brown of Capital Economics noted that Information and Communication Technology (ICT) output rose in the third quarter despite drops in employment, further implying that AI is making a large, tangible contribution to this growth [1].
The Decoupling of Labor and Output
Recent government data supports the narrative that economic output is decoupling from labor intensity. A report released on Thursday, February 12, 2026, significantly revised 2025 job gains downward to 181,000, a steep drop from the initial estimate of 584,000 and the 1.46 million jobs added in 2024 [1]. Despite this reduction in labor input, real gross domestic product (GDP) remained robust, posting 3.7% growth in the fourth quarter [2]. Economists interpret this combination of lower employment growth alongside strong economic output as a classic indicator of rising productivity [2][3].
From Experimentation to Structural Utility
This divergence in the labor market underscores what Brynjolfsson describes as a transition “from an era of AI experimentation to one of structural utility” [1][2]. The focus has shifted from testing the capabilities of generative models to integrating them into the core mechanics of business operations. While Torsten Slok, Chief Economist at Apollo, previously noted that AI was absent from macroeconomic data, the revised figures for 2025 suggest the technology is finally showing up in national statistics [1][3]. However, experts caution that this trend is not immune to external shocks; Brynjolfsson warns that the productivity revival could still be offset by geopolitical tensions or monetary policy errors [1][2].