AI Effects on Firms and Workers: Brookings Research on Growth, Jobs, and Innovation

Understanding AI Investment and Firm-Level Impact

The past decade has witnessed a transformative expansion of commercial artificial intelligence investments, reshaping how businesses operate across every sector of the global economy. From the pivotal 2012 ImageNet challenge that catalyzed breakthroughs in computer vision and deep learning to the release of ChatGPT in late 2022, AI adoption has accelerated at an unprecedented pace. A landmark study by Babina, Fedyk, He, and Hodson (2024), published in the Journal of Financial Economics, offers one of the most comprehensive analyses of how these investments translate into real-world business outcomes.

The research leverages detailed employer-employee data covering as much as 64% of the US workforce to track individual companies’ AI investments and their accompanying effects on operations and employment. Unlike previous studies that focused on occupation-level or industry-level effects, this firm-level approach captures not just AI producers but all firms using artificial intelligence in their everyday operations. As organizations increasingly integrate AI into their workflows, understanding these dynamics is critical for business leaders, policymakers, and workers navigating the evolving landscape of AI-driven economic transformation.

The methodology developed by the researchers provides a novel way to measure firm-level AI investments. By tracking AI researchers and software engineers through job postings and resume data, and assigning AI-relatedness scores to required skills, the approach captures the full spectrum of commercial AI adoption. Skills like TensorFlow receive high AI-relatedness scores (0.9), while unrelated skills score zero. This granular tracking reveals that AI workers, while highly specialized and representing roughly 1% of firms with dedicated AI teams, serve as reliable indicators of a company’s commitment to artificial intelligence technologies.

AI Spurs Firm Growth and Increases Employment

Perhaps the most significant finding from the Brookings research challenges the widespread fear that AI adoption inevitably leads to job displacement. The data tell a remarkably different story: firms that invested more heavily in artificial intelligence actually experienced substantial growth in both sales and employment. A one-standard-deviation difference in AI investments translated into approximately 20% higher sales growth over a decade, equivalent to roughly 2% additional annual growth.

Equally important, this sales growth was accompanied by proportional employment increases. Firms with greater AI investments expanded their total employee headcount at rates mirroring their revenue growth. Both costs of goods sold and operating expenses increased proportionally to sales, suggesting that AI-investing firms were genuinely expanding rather than simply automating away their workforce. This finding stands in sharp contrast to public discourse that often frames AI as a direct threat to employment.

The timing of these effects provides critical context for business planning. Benefits from AI investments typically do not appear immediately but take approximately two to three years to materialize in sales growth, after which the increase becomes persistent. This lag is consistent with historical patterns of technology adoption described by Brynjolfsson, Rock, and Syverson (2019), who note that firms require time to develop complementary assets and organizational capabilities to fully leverage new technologies. Companies considering AI investments should therefore adopt a patient, long-term perspective rather than expecting immediate returns.

The Productivity Paradox of Artificial Intelligence

Despite the strong growth in sales and employment, a curious finding emerges: aggregate productivity measures have not moved significantly over the past decade of accelerating AI investments. Several academic studies, including Rock (2019) and Babina et al. (2024), find that AI investments have not been associated with increases in either sales per worker or revenue total factor productivity. This phenomenon echoes the broader productivity paradox first identified in the context of earlier digital technologies.

This finding does not necessarily mean AI fails to improve efficiency. Rather, it suggests that the primary mechanism through which AI benefits firms is growth and expansion rather than cost reduction. In most sectors, companies are using AI to create new products, enter new markets, and serve more customers rather than simply replacing existing workers with automated systems. The growth in both revenue and headcount at roughly similar rates means that per-worker productivity metrics remain flat even as total output increases substantially.

There are important exceptions to this general pattern. In sectors where AI is particularly well-suited to the task and where growth potential is limited, labor-displacement effects can be significant. For example, research by Fedyk et al. (2022) shows that audit firms using AI have reduced their audit workforce. Similarly, Brynjolfsson et al. (2025) find that AI tools dramatically increase customer service worker efficiency. But these sector-specific effects do not dominate the aggregate picture, where growth and expansion remain the primary channels through which AI impacts firm performance.

AI-Fueled Growth Through Product Innovation

The research reveals that AI-fueled growth is overwhelmingly driven by product innovation rather than process optimization. Over the 2010-2018 period, a one-standard-deviation increase in firm-level AI investments was associated with a striking 24% increase in product patents and a 13% increase in trademarks. In contrast, process patents increased by just over 1%, a statistically insignificant effect. This pattern strongly suggests that companies are deploying AI primarily to innovate and create rather than to streamline and cut costs.

AI-powered innovation spans a wide spectrum, from incremental improvements to breakthrough discoveries. Computer vision technologies that help autonomous vehicles detect obstacles represent incremental quality improvements that make existing products better and safer. On the other end of the spectrum, the leadership of Moderna has credited machine learning and AI as the driving force behind their ability to rapidly develop the COVID-19 vaccine, compressing experimentation timelines from years to months. These examples illustrate how AI serves as a general-purpose technology that enhances innovation across fundamentally different domains and applications.

For organizations evaluating their AI strategies, this finding has significant implications. Firms that focus exclusively on using AI for cost reduction and process automation may be missing the larger opportunity. The evidence suggests that the greatest returns from AI investments come through product development, market expansion, and the creation of entirely new offerings. This innovation-first approach explains why AI-investing firms grow their workforces rather than shrinking them—new products and services require new teams to develop, market, and support them.

Workforce Upskilling and the AI Skills Gap

While the overall relationship between AI adoption and employment is positive, the composition of firms’ workforces undergoes significant transformation. Research by Babina et al. (2023) demonstrates that AI-investing firms systematically shift their hiring toward more educated and technically skilled workers. Over eight years, a one-standard-deviation increase in AI investment was associated with a 3.7% increase in college-educated workers, a 2.9% increase in masters degree holders, and a 0.6% increase in doctoral degree holders. Correspondingly, the share of workers without college degrees declined by 7.2%.

Beyond education levels, AI firms also shift the types of degrees they value. The share of employees with STEM backgrounds increases while the relative share of social science, arts, and other non-technical majors declines. This skill-biased technological change mirrors patterns observed with previous waves of technology adoption, as documented extensively by labor economists including Autor, Katz, and Krueger (1998) and Katz and Murphy (1992). AI, like computing before it, favors workers with higher technical capabilities.

Since total employment at AI-investing firms increased, these compositional shifts do not necessarily mean that less-educated workers were fired. Rather, the reallocation occurs primarily through new hiring patterns, with AI firms seeking an increasingly skilled workforce for their expanding operations. This distinction is important for policy: the challenge is not mass unemployment but rather ensuring that workers can acquire the skills needed to participate in the AI-driven economy. Reskilling and continuous education programs become essential for workforce adaptation, a theme explored in detail in our analysis of OECD strategies for building an AI-ready public workforce.

How AI Reshapes Corporate Hierarchies

One of the more fascinating findings from the research concerns how AI investments transform corporate organizational structures. The data reveal a clear flattening of hierarchies at AI-investing firms. A one-standard-deviation increase in AI investments from 2010 to 2018 was associated with a 1.6% increase in the share of junior employees—independent contributors not managing others—while middle management decreased by 0.8% and senior management by 0.7%.

This empirical finding resolves an important theoretical ambiguity. On one hand, the increased complexity from AI-driven innovation might require more management layers. On the other hand, as Garicano and Rossi-Hansberg (2006) predicted, technology that improves knowledge acquisition can increase individual employees’ span of control, reducing reliance on hierarchical structures. The data support the latter hypothesis: AI empowers individual workers to access information and solve problems that previously required escalation through management chains.

Importantly, this flattening trend was not a general phenomenon across all US firms during the study period. The shares of junior employees, middle management, and senior management remained relatively stable across US public firms from 2010 to 2018. The shift toward flatter structures was unique to AI-investing firms, suggesting a direct causal link between AI technology adoption and organizational restructuring. For corporate leaders, this implies that successful AI integration requires rethinking not just technology systems but entire management philosophies, empowering skilled individual contributors with greater autonomy and decision-making authority.

AI Effects on US Industries and Market Concentration

Beyond firm-level effects, the research examines how AI investments reshape entire industries. At the aggregate level, both industry-level sales and employment increase with AI adoption, at least among publicly traded firms. This means AI creates genuine value rather than simply redistributing market share from non-adopters to adopters. Looking at total industry employment including private firms shows milder but still positive growth, with some reallocation from smaller private companies to larger public ones.

This finding contrasts notably with the effects of previous automation technologies. Research by Acemoglu, Lelarge, and Restrepo (2020) found that robotics investments led to firm-level employment growth but industry-level employment decline—a classic reallocation effect where automated firms grew at the expense of competitors, reducing total jobs. AI, at least so far, does not exhibit this displacement pattern at the industry level, suggesting it functions more as a rising tide that lifts most boats rather than a zero-sum competitive weapon.

However, the distributional effects within industries tell a more nuanced story. AI investments are associated with increased industry concentration, measured both by top-firm market share and the Herfindahl-Hirschman Index. This concentration trend raises important questions about long-term market competitiveness and consumer welfare. While AI-investing firms have not yet increased price markups, it remains plausible that firms may eventually leverage their AI-enhanced market dominance to raise prices, as explored in broader analyses of how AI is reshaping global trade and competitive dynamics.

Small Firms vs. Large Firms in the AI Economy

The research reveals a stark divergence in AI benefits based on firm size. When companies are divided into thirds by initial size, the effects of AI investments are most pronounced for the largest firms, moderate but significant for mid-sized firms, and statistically insignificant for the smallest firms. Among smaller companies, there was virtually no difference between those that invested in AI and those that did not.

This size-dependent effect has straightforward explanations. Larger firms possess extensive proprietary data repositories that serve as fuel for AI models, dedicated technical teams with the expertise to implement sophisticated systems, and the financial resources to sustain the two-to-three-year investment horizon before returns materialize. Smaller firms typically lack these complementary assets, making it difficult to extract meaningful value from AI investments even when they have access to similar technologies.

The implications for economic policy are significant. Without intervention, AI could accelerate the already pronounced trend toward corporate consolidation, with large incumbents using their AI advantages to entrench market positions. Evidence from Babina et al. (2025) suggests that policies promoting data access—such as open banking frameworks that allow customers to share their financial data with fintech competitors—can help level the playing field. By reducing the proprietary data advantage of large incumbents, such policies enable smaller, innovative firms to compete more effectively in the AI economy.

Policy Implications for the AI-Driven Economy

The rapid diffusion of AI across firms has begun reshaping labor markets, organizational structures, and industry dynamics in ways that demand thoughtful policy responses. While the evidence is largely encouraging—AI drives growth, creates employment, and fuels innovation—the benefits accrue disproportionately to larger firms and more highly skilled workers. Policymakers face the challenge of nurturing AI-driven growth while ensuring its benefits are broadly shared across the economy.

Education and workforce development emerge as the most critical policy priorities. The strong skill-bias in AI-investing firms’ hiring patterns means that workers without college degrees or STEM skills face increasing competitive disadvantages. Investing in STEM education, expanding access to digital skills training, and creating robust mid-career reskilling programs are essential to prevent a widening skills gap from undermining labor market equity. These initiatives must keep pace with AI’s rapid evolution, requiring continuous curriculum updates and industry-academia partnerships.

Antitrust and competition policy also requires adaptation. The finding that AI increases industry concentration without yet raising prices creates a complex regulatory challenge. Traditional antitrust frameworks focused on consumer harm through pricing may need to evolve to consider concentration risks from AI-driven market dominance. Additionally, expanding access to data through open data frameworks can help democratize AI’s benefits, enabling smaller firms to compete on more equal footing with data-rich incumbents.

Lessons for Business Leaders and Workforce Planning

For business leaders navigating the AI-driven economy, the Brookings research offers several actionable insights. First, AI investments should be viewed as long-term growth strategies rather than short-term cost-cutting measures. The evidence overwhelmingly shows that AI’s primary value lies in product innovation and market expansion, not workforce reduction. Companies that approach AI with an innovation-first mindset are likely to realize the greatest returns.

Second, successful AI adoption requires concurrent investment in human capital. The shift toward more educated and technically skilled workforces means that training and development programs are not optional supplements but essential complements to technology investments. Organizations should invest in upskilling existing employees, recruiting STEM talent, and creating organizational cultures that encourage continuous learning and adaptation.

Third, the flattening of corporate hierarchies observed at AI-investing firms suggests that leadership models must evolve. Empowering individual contributors with greater autonomy, reducing bureaucratic layers, and fostering a culture of experimentation align with the organizational changes that AI naturally induces. Companies that resist these structural shifts may find themselves unable to fully capitalize on their AI investments, regardless of how sophisticated their technology deployments become.

Finally, smaller firms should not be discouraged by the size-dependent effects identified in the research. While the challenges are real, the same study points to pathways for smaller organizations to compete effectively. Focusing AI investments on specific, high-impact use cases rather than broad deployments, leveraging cloud-based AI services that reduce the need for proprietary infrastructure, and participating in data-sharing ecosystems can help smaller firms extract meaningful value from artificial intelligence technologies.