The AI Revolution Won’t Transform the Economy Unless Small Businesses Are Included

Artificial intelligence (AI) is often described as the next general-purpose technology, holding the promise of transforming economies much as electricity and the internet once did. That promise, however, depends not on technological breakthroughs alone, but on whether AI adoption spreads widely across firms, sectors, and regions. Small and medium-sized enterprises (SMEs)—which account for more than 90 percent of businesses worldwide and nearly half of global GDP—sit at the center of this economic transition.

For most SMEs today, AI is experienced not as advanced machine learning systems, but as generative and analytical tools embedded in everyday software. More transformative applications remain limited, as adoption is uneven-shaped by high costs, limited technical capacity, constrained access to computing infrastructure, and rising data and regulatory concerns. These challenges are not purely technological; they reflect broader gaps in digital readiness, institutional support, and market access that disproportionately affect smaller firms.

SMEs in the AI Value Chain: Upstream and Downstream Roles

Examining national AI and digital strategies across 80 countries, the findings show that SMEs play both upstream and downstream roles in the AI value chain. National strategies address these constituencies in different ways—sometimes explicitly as distinct policy targets, and at other times implicitly alongside startups and other enterprises.

Upstream SMEs, or solution designers alongside startups, are supported through mechanisms such as research and development grants, innovation financing, regulatory sandboxes, test beds, and preferential public procurement. More commonly, national strategies also emphasize SMEs’ downstream role as adopters of AI technologies, using tools such as digital maturity assessments, AI training, innovation hubs, and advisory services to reduce adoption barriers.

History shows that technological revolutions become truly transformative only when diffusion is broad. AI’s potential lies not in technical sophistication, but in how widely its productivity gains spread across the economy. Like electricity and computers—which became commercially viable decades before generating economy-wide productivity gains—AI today appears to be in a similarly early stage of diffusion.

Several national strategies explicitly stress that the largest economic gains from AI are expected to come from diffusion rather than frontier innovation alone. Ireland for example, highlights that the greatest economic benefit arises when enterprises of all sizes adopt digital technologies. Germany emphasizes transferring AI capabilities into an economy dominated by small and medium-sized enterprises—the Mittelstand. Singapore’s “SMEs Go Digital” program similarly prioritizes adoption across multiple sectors.

Digital Readiness and Infrastructure Constraints

In several countries, AI is embedded within broader digital transformation agendas that could predate dedicated AI frameworks, reflecting a growing recognition that digital readiness is a prerequisite for effective adoption. Armenia, for example, illustrates this sequencing, with SME digitalization efforts focused on connectivity, cloud adoption and workforce skills emerging before formal AI initiatives.

Access to computing infrastructure remains a major constraint for SMEs, as cloud services and high-performance computing are often prohibitively expensive. Governments have responded with shared cloud platforms, computing credits, and targeted financial instruments—such as AI voucher schemes in South Korea and the Netherlands, and Thailand’s Financial AI initiative, which uses alternative data to expand SME access to technology finance.

The U.S. Contrast

U.S. AI strategies present a notable contrast, with only implicit references to SMEs alongside “industry,” “the private sector,” and “startups.” Although recent strategies acknowledge that advanced AI infrastructure can be prohibitively expensive for smaller firms, concrete diffusion mechanisms remain largely absent.

The divergence across countries is not accidental—it reflects the economic DNA shaping how innovation and adoption are organized. The U.S., a liberal market economy, prioritizes competition and private investment, making it exceptionally strong at frontier innovations but weaker at ensuring technologies spread widely. Coordinated market economies such as Germany and Singapore, by contrast, rely on institutions and policy coordination to translate innovation into broad-based adoption.

These institutional differences are reflected in economic outcomes. Studies show that the productivity gap between frontier and lagging firms has widened far more in the United States than in most European economies. While leading firms in both regions have grown rapidly, only coordinated economies have generated meaningful productivity spillovers to small and mid-sized enterprises.

What emerges is that AI’s economic impact depends less on technological capability than on institutional design and diffusion capacity. Small and medium-sized enterprises are not peripheral users waiting for innovation to trickle down; they are central actors in both the creation and adoption of AI-driven productivity. Where governments align digital readiness, infrastructure access, skills development, and financing through coordinated diffusion strategies, AI can fast-track the trajectory of earlier technological revolutions—delivering economy-wide gains far more quickly than past general-purpose technologies. Where such coordination is absent—particularly in market-driven systems that privilege frontier innovation over broad adoption—AI risks deepening existing divides between leading firms and the rest of the economy. Ultimately, the trajectory of the AI revolution will be shaped not by breakthroughs at the frontier, but by whether public policy deliberately enables smaller firms to participate, adapt, and scale alongside it.