Harvesting the AI Dividend
Productivity, typically measured as output per hour worked, is the primary long-term driver of income growth and living standards. Both the U.S. and Europe have experienced slower productivity growth since the mid-2000s compared with earlier decades.
Now, however, many economists and policymakers view AI as a potential catalyst for reversing that slowdown. AI—especially the rise of generative AI and AI agents—is widely expected to shape the next phase of productivity growth in advanced economies, including those in the U.S. and Europe.
The key question for business leaders is not whether AI will matter, but how large the productivity gains will be, how quickly they will materialize, and which region will benefit most.
Productivity growth
The Organization for Economic Co-operation and Development (OECD) estimates that AI could raise annual labor productivity growth in advanced economies by roughly 0.4 to 1.3 percentage points, depending on adoption intensity and sector exposure. These gains would be meaningful because even an additional half percentage point of annual productivity growth compounds significantly over a decade.
However, the OECD and other economists stress that outcomes depend heavily on complementary investments in digital infrastructure, workforce training, and organizational change, rather than on technology alone.
Between 1995 and 2019, U.S. labor productivity grew at 2.1% annually compared to one percent in Europe. This disparity arose in part because companies in the U.S. invested more aggressively in information, communications, and technology while those in Europe were constrained more by regulatory and other factors.
Expectations for AI-driven productivity gains remain generally stronger in the U.S. than in Europe. Goldman Sachs suggests that widespread adoption of generative AI could raise U.S. labor productivity growth by around one to 1.5 percentage points per year.
Several structural factors support this view. The U.S. has a deep technology ecosystem, global leadership in AI research and venture capital, and a large, digitally intensive services sector, including finance, professional services, and IT, where generative AI tools can be rapidly deployed.
Agentic AI
In both Europe and the U.S., AI agents represent a particularly important development. Unlike earlier automation tools that handled isolated tasks, AI agents—like Joule Agents from SAP—are designed to plan, reason, and execute multi-step workflows. For example, an agent might manage customer service tickets, draft responses, query databases, escalate issues, and update systems—all with limited intervention.
In knowledge-based industries, this kind of workflow automation could significantly raise output per worker. But rather than replacing entire occupations, AI agents may reduce time spent on repetitive administrative and “long-tail” tasks, enabling workers to focus on higher-value analysis, strategy, and interpersonal activities.
Despite stories about failed corporate AI projects, which can typically involve bolt-on or stand-alone AI pilots rather than a more integrated, holistic approach, recent evidence from the U.S. suggests that productivity gains are already emerging in some sectors. For example, financial institutions have reported significant efficiency improvements in back-office operations through AI deployment.
Similarly, experimental studies in professional services show that generative AI can increase output quality and speed, particularly for less experienced workers, effectively narrowing skill gaps within teams.
European outlook
The outlook for productivity gains in Europe from AI is more mixed. According to a recent International Monetary Fund (IMF) report the medium-term gain in productivity from the AI alone would vary considerably across countries, and for Europe as a whole would be rather modest: about 1.1 percent cumulatively over five years.
But with pro-growth reforms, the IMF suggests that much bigger gains are possible over the longer run. Like the OECD, the IMF emphasizes that regulatory frameworks, labor market structures, and the pace of technology diffusion will strongly influence outcomes.
Several structural differences shape Europe’s trajectory and the size of what has been called the “AI growth dividend.” First, AI adoption among small and midsize enterprises (SMEs), which form a larger share of the European economy than in the U.S., tends to be slower. Second, Europe’s digital market remains more fragmented across national boundaries, languages, and regulatory systems, which can complicate scaling technology platforms. Third, the European Union has taken a more precautionary regulatory approach to AI governance. While this may reduce certain risks, it could also dampen short-term productivity gains if compliance burdens slow deployment.
Europe’s strengths
That said, Europe has strengths. It leads in advanced manufacturing and industrial engineering, sectors where AI-driven optimization, robotics, and predictive maintenance can raise capital productivity. In these areas, AI agents embedded in industrial systems could significantly enhance supply chain efficiency and reduce downtime.
In addition, as SAP executives have pointed out, Europe has an enormous repository of structured business and manufacturing data, which is essential for reliable and effective AI systems as well as trust in AI Agents.
If AI adoption accelerates in manufacturing and energy systems and if European companies seize the opportunity to build advanced AI agents and apps using their business data, Europe could see much more robust medium-term productivity gains. As an example, SAP’s internal use of AI tools has already significantly improved its own developer productivity.
Labor flexibility
A critical factor in both the U.S. and Europe is labor market adjustment. Historically, the U.S. labor market has demonstrated greater flexibility, with higher rates of job switching and occupational mobility. This flexibility may facilitate faster reallocation of workers into AI-complementary roles, amplifying productivity gains, though this could be offset by more effective existing workforce retraining.
As the Bank for International Settlements (BIS) has noted, AI’s productivity effects are unlikely to be automatic. Productivity gains from AI depend on complementary investments in skills, management practices, and digital infrastructure. The BIS warns that without these, AI tools may produce only marginal efficiency improvements.
The historical lesson from past general-purpose technologies, such as electricity and IT, is that productivity surges occur only after organizations redesign processes to exploit new capabilities and take a holistic rather than piecemeal approach toward implementation.
No AI bubble
While some investors have expressed concerns about an AI bubble, total AI spending in the U.S. is still below one percent of GDP. Joseph Briggs, senior global economist at Goldman Sachs, notes that this is well below historical infrastructure cycles. For comparison historical infrastructure investments such as IT spending, railroads and canals typically represented between two and five percent of GDP.
Like these previous investment waves AI, particularly agentic AI, is likely to generate significant productivity growth and a corresponding boost to GDP in those regions and sectors that seize the AI opportunity.
