Rodney Sullivan, Executive Director of the Mayo Center of Asset Management at the Darden School of Business (University of Virginia), publishes an analysis on the emergence of a productivity boom in the United States, which he primarily attributes to the adoption of artificial intelligence in work processes.
What the macroeconomic data shows
By the fourth quarter of 2025, the real American GDP growth is estimated at 4.2%, driven mainly by a significant increase in private domestic investment. This performance is accompanied by a structural peculiarity: it does not rely on strong job creation. The labor force is growing slowly due to demographic dynamics, a slowdown in migration flows, and the constraints of returning to the labor market post-COVID. As a result, production is increasing faster than labor input, which precisely defines an increase in productivity.
Technology companies have invested about $180 billion more than the previous year in software and research and development. Unlike previous waves of IT investment, including the Internet revolution of the 1990s, which mainly improved information circulation, current AI systems directly enhance cognitive and operational tasks: drafting initial versions, generating code, logistic optimization, customer support. Economically, AI functions as a form of evolutionary work, allowing a given worker to supervise and deploy a much larger production than before.
Why this is not (yet) an employment shock
Rodney Sullivan notes that the rise in productivity has not yet been accompanied by widespread labor market tensions. Layoffs remain contained, and unemployment claims are moderate. Workers are shifting towards supervisory, coordinating, and judgment roles, while machines take over routine tasks. The transition is smooth: machines complement human work rather than massively replacing it, at least at this stage.
This dynamic has constructive macroeconomic implications. Productivity-led growth alleviates inflationary pressures by limiting unit labor costs, allowing real incomes to increase without prompting the Federal Reserve to raise rates.
Implications for investors
When productivity growth surpasses wage growth, unit labor costs moderate – which is the most benign form of disinflation: prices stabilize because supply becomes more efficient, not because demand collapses. For companies, this mechanism supports margin expansion through operational leverage, especially in non-labor-intensive sectors.
Sullivan points out that productivity gains are more durable than those from temporary pricing power, and they are more evenly distributed across sectors. This provides a stronger foundation for stock valuations than a technology boom focused on mega-capitalizations. Productivity gains linked to AI are expected to increasingly benefit companies using the technology, not just those creating it, an argument for broadening performance beyond the narrow circle of large tech companies.
The paradox of the early career employment market
Sullivan identifies an counter-intuitive tension: strong GDP growth coexists with a more challenging entry-level job market. When companies can increase their production through technology rather than expanding their workforce, recruitment becomes more selective. Entry-level positions – which served as a learning ground for tasks like drafting, data cleaning, basic analysis – are precisely where AI excels. The premium now goes to profiles capable of exercising judgment, integrating information, and supervising automated output.
Sullivan
“These are the first steps of an AI-driven growth regime. Not as a sci-fi scenario, not as an employment apocalypse, but for now, as a set of tools enabling companies to extract more value from the workers they already have.”
