Does the study prove that AI creates jobs?

No. It shows that AI users are more likely to expect growth. Causal proof and the firms' later outcomes are still missing.

How many businesses were studied?

The AI module covered 5,248 US firms with at least one employee; the adjusted subsamples were slightly smaller.

Have the firms already earned more revenue?

Thirty-one percent reported higher sales from AI. The strongest findings, however, concern expectations for the following twelve months.

Does the result apply to Germany?

Not automatically. Regulation, industry structure and labor markets differ, so separate European evidence is needed.

AI at small firms: stronger job and revenue expectations

What this is about

An analysis published on October 8, 2026 by the Federal Reserve Bank of New York challenges a common assumption: Small businesses using AI are not more likely to expect job cuts. They are considerably more likely to expect employment and revenue growth. The evidence comes from the 2025 Small Business Credit Survey, whose AI module covered 5,248 US employer firms.

This matters because arguments about AI and work often focus on large corporations or individual occupations. Small firms have less capital and fewer technical employees, yet account for a large share of employment. Their expectations offer a view of how AI may affect work beyond the best-known platform companies.

What the analysis actually does

Researchers Will Aarons and Asani Sarkar compare firms already using AI or planning to adopt it with non-users. Forty-six percent of surveyed employer firms used AI, and another 15 percent planned to adopt it within twelve months. Among users, 63 percent described AI as at least somewhat important to production, while 51 percent had partly or fully integrated it into business processes.

For employment, the net share of AI users expecting growth rather than decline was 33 percentage points. The equivalent figure for non-users was 15 points. After accounting for firm, owner and location characteristics as well as employment changes during the previous year, a 10-point gap remained. For revenue expectations, the adjusted advantage was 14 percentage points.

Why it matters

The figures conflict with a simple story in which AI immediately replaces labor at small firms. Seventy-seven percent reported no current change in labor costs from AI. Only 31 percent said AI had already increased sales. The optimism therefore concerns the future more than gains already measured.

One result stands out: Among AI users, firms facing operational technology problems had net revenue-growth expectations 16 points above AI users without those challenges. This supports the idea that accessible tools could ease bottlenecks in marketing, ecommerce or customer service. Research from Stanford, Yale and the Federal Reserve Board nevertheless shows a mixed labor-market picture, particularly for early-career workers and occupations highly exposed to AI.

In plain language

Think of AI as an extra mixer in a small bakery. It can mix dough faster and sort orders, but it cannot decide whether enough customers will arrive or whether the bakery will hire another person. A bakery buying the mixer may already expect to grow. That is why the relationship is interesting but does not prove that the machine caused the growth.

A practical example

Consider a fictional online retailer with eight employees. The team uses AI to sort 120 customer requests a day, draft product descriptions and flag unstable inventory levels. The owner consequently plans to add two sales jobs next year instead of hiring one person to handle routine emails.

In the survey, this firm would appear as an AI user with positive employment and revenue expectations. The study has not yet measured whether twelve months later the company actually employs ten people or earns more revenue.

Scope and limits

First, the study measures expectations, not outcomes that later occurred. Owners can be wrong. Second, it is observational: Growth-oriented firms may be more inclined to adopt AI, meaning adoption is not necessarily the cause of optimism. Statistical controls reduce this concern but cannot eliminate it. Third, the data come from US firms surveyed between September and November 2025; they cannot be transferred directly to Germany, larger companies or the tool landscape of late 2026.

The 31 percent reporting increased sales also does not mean AI alone produced that revenue. Stronger evidence will require follow-up measurements that compare each firm's expectations with its actual trajectory.

SEO & GEO keywords

Federal Reserve Bank of New York, Small Business Credit Survey, AI and jobs, small businesses, employment growth, revenue growth, AI adoption, US labor market, automation, Will Aarons, Asani Sarkar

Small firms using AI are more likely to expect job growth