How Is AI Impacting Small Business Performance? Evidence from the Third District

As adoption of artificial intelligence (AI) applications rises among U.S. firms, researchers are seeking to better understand the technology's usage and impact across different types of businesses. Recent research suggests that although large businesses currently exhibit higher rates of AI adoption than small businesses, the latter are projected to narrow this gap by the end of 2026. Understanding how small businesses implement and benefit from this technology is particularly important because they employ roughly half of U.S. workers. Additionally, small firms face greater financial constraints and typically lack the robust information system infrastructure of larger firms.

Using data from the 2025 Small Business Credit Survey (SBCS) special question module¹ on artificial intelligence, we explore small business AI adoption, task-specific usage, and performance impacts in Third District states (Delaware, New Jersey, and Pennsylvania).² We find that AI adoption among Third District firms lags the national rate by nearly 10 percentage points at 37 percent, potentially reflecting differences in industry composition. Among adopters in the Third District, the most common uses are writing and marketing (79 percent), individual productivity tasks (e.g., notetaking, summarizing) (56 percent), and planning or analysis (e.g., research) (51 percent). Over 70 percent of firms reported no change in core business operations, including labor costs, spending on outside services, production quality, and sales, although 71 percent reported increases in productivity.

Third District Small Businesses Have Lower AI Adoption Rates Compared with the Nation

At the time of the survey in fall of 2025, a smaller share of Third District firms reported using AI compared with all U.S. firms: 37 percent of firms in the Third District versus 46 percent in the nation (Figure 1). Of Third District AI adopters, the majority described their level of AI adoption as "experimenting with the technology" (57 percent), and few had fully integrated AI into their business operations (3 percent). Almost half of Third District firms reported not using AI, with no plans for adoption, considerably higher than the one-third of businesses reporting the same nationwide.

One explanation for the lower regional AI adoption intensity is that Third District firms tended to see the technology as less important to their core business functions than U.S. firms. Forty-four percent of Third District firms using AI reported that AI was not important to their business's core goods and services, compared with 37 percent of U.S. firms (Figure 2). This discrepancy could be driven by differences in industry composition: The Third District sample has more manufacturing (12 percent in the District versus 4 percent in the nation) and less nonmanufacturing goods production (15 percent versus 20 percent) and services (73 percent versus 76 percent). As our report last year showed, production workers had among the lowest generative AI exposure scores of major occupation groups, suggesting that manufacturing firms may have fewer use cases for generative AI than service firms.

Majority of Regional Firms Use AI for Writing or Marketing, Individual Productivity, and Planning or Analysis

Figure 3 shows the share of small businesses in the Third District that use AI by specific processes.³ Nearly 80 percent of Third District firms said they use AI for writing or marketing, while over half reported using it for individual productivity and planning or analysis. Roughly one-third of firms cited using AI for administrative business functions and customer service, 16 percent for process automation, and only 12 percent for coding-related tasks.

Third District Firms Reported No Changes Across Most Performance Metrics Due to AI

Most Third District firms using AI reported no changes across their performance metrics due to the technology, with the exception of productivity. Over 70 percent of firms saw no change in labor costs, spending on outside services, quality of goods and/or services, or sales due to AI. In contrast, 71 percent of firms reported AI-related increases in productivity, although the survey did not ask how firms measured productivity gains. Despite these self-reported increases in productivity, few Third District firms reported decreases in labor costs or spending on outside services resulting from AI usage, implying that AI had not significantly reduced regional small businesses' spending on internal or external labor at the time of the survey. When the national SBCS sample⁴ was asked about employment changes due to AI, 79 percent of firms reported no change, 11 percent reported decreases, and 10 percent reported increases.

Productivity Gains Without Substantial Job Losses - So Far

A common concern among workers and policymakers is that AI-driven productivity gains will lead to major job losses. However, results from the Small Business Credit Survey suggest otherwise: Although most firms experienced productivity gains, few reported decreasing staff or labor costs. Similarly, another survey of both small and large Third District businesses found that AI adoption has had little impact on firm demand for workers. Research from the New York Fed indicates that more Second District firms are retraining workers rather than reducing labor due to AI usage. Overall, empirical evidence on AI adoption and labor demand remains mixed, with no clear causal link between the technology and job losses - either in AI-exposed occupations or in junior hiring.

Research from OpenAI explains that higher worker productivity due to AI does not necessarily mean fewer jobs. Other factors, such as demand elasticity (i.e., how much demand changes when the price for a good or service changes due to AI-driven productivity gains) and the human necessity of specific roles, will determine future labor demand across occupations as the technology becomes more widely adopted. However, a recent working paper surveying over 6,000 executives across the United States, the United Kingdom, Germany, and Australia presents a more cautious outlook: Although AI has had minimal effects to date, executives anticipate significant job losses and productivity gains over the next three years.

The national sample for the Small Business Credit Survey found that over 20 percent of firms using AI cite challenges such as accuracy, adapting the technology to business needs, data security and privacy concerns, training time, and costs, all of which could create barriers for small businesses considering AI adoption. As the landscape evolves, future generative AI adoption rates may be influenced by changing AI computing costs and competitive pressures between firms that adopt AI and those that don't. Continued monitoring of adoption patterns and employment outcomes will be essential to understanding the technology's long-term impact on regional firms and labor markets.