Field notesUpdated July 2026

Small-business AI adoption: the statistics

Three numbers frame it. 76% of small businesses now use AI, per Goldman Sachs' 10,000 Small Businesses Voices survey, 2026, yet only 14% have it fully embedded in core operations, per the same study. And about half of AI-using small firms have invested nothing in training or integration, per the SBA Office of Advocacy, 2025.

This page collects every statistic I actually use when talking with owners about AI: how many small businesses have adopted it, how few have integrated it, and where the gap actually lives. Each number carries its source, its vintage, and a link to the primary. Last revised July 2026.

No pitch. You'll leave with two or three concrete next steps, whether or not we work together.

By Nick Jones, Founder and Operating Partner. Cite freely; every number links to its primary source. More about Nick

76%

of small businesses now use AI in some form

Goldman Sachs 10,000 Small Businesses Voices, 2026

14%

have AI fully embedded in their core operations

Goldman Sachs, 2026

68%

use AI regularly, up from 48% in July 2024

QuickBooks-commissioned survey, April 2025

~50%

of AI-using small firms have invested nothing in training or integration

SBA Office of Advocacy, 2025

Integrity note

How to read these numbers

Every statistic on this page is labeled with its source and its vintage. Survey numbers are self-reported, and I say so rather than dress sentiment up as measured ROI. Where the strongest evidence comes from enterprises instead of small businesses, as with MIT's pilot-failure data below, it's labeled as an enterprise finding instead of being passed off as a small-business one. Links go to primary sources wherever a primary exists. Sources revise their work, so when they update, this page updates, and the date at the top changes with it. If you spot a number that has drifted from its source, email nick@eatsleeplaunchrepeat.com and I'll fix it.

Adoption

AI adoption is mainstream among small businesses.

By 2026, a clear majority of US small businesses use AI in some form: 76% per Goldman Sachs, 2026, 68% per a QuickBooks-commissioned survey, 2025, and 58% for generative AI specifically, per the US Chamber of Commerce, 2025. Three surveys, three definitions, one direction. The adoption question is settled. The integration question, covered below, is not.

76% of small businesses use AI

And 93% of users report a positive impact, per Goldman Sachs' 10,000 Small Businesses Voices survey (n=1,256, fielded January 27 to February 4, 2026).

Goldman Sachs 10,000 Small Businesses Voices, 2026
68% use AI regularly

Up from 48% in July 2024, with 28% using it daily. 74% of AI users say it makes them more productive, up from 46%. Per a QuickBooks-commissioned survey of 2,200+ US businesses with up to 100 employees, fielded April 2025.

QuickBooks-commissioned survey, April 2025
58% use generative AI

Up from 40% in 2024 and 23% in 2023, per the US Chamber of Commerce and Teneo, August 2025. In the same study, 82% of AI-using small businesses grew their workforce, and generative AI ranked in the top three at-work technologies, behind search engines and ahead of social media.

US Chamber of Commerce / Teneo, August 2025

The integration gap

The gap is integration, not adoption.

Usage is broad; embedding is rare. Only 14% of small businesses have AI fully embedded in core operations, per Goldman Sachs, 2026, and about half of AI-using small firms have invested nothing in training or integration, per the SBA Office of Advocacy, 2025. Tools got adopted. Workflows, mostly, did not change.

Only 14% have AI fully embedded

76% of small businesses use AI, but just 14% have it fully embedded in core operations, and 73% want more training and implementation resources, per Goldman Sachs, 2026. The distance between those numbers is where the leverage is.

Goldman Sachs 10,000 Small Businesses Voices, 2026
~50% have invested nothing in training or integration

About half of AI-using small firms have spent nothing on training for, or integration of, the tools they already use, per the SBA Office of Advocacy's analysis of Census Bureau BTOS data, September 2025.

SBA Office of Advocacy / Census BTOS, September 2025
82% say AI adoption is essential to compete

Practical training is the number one support need, and 74% want clearer ROI evidence, per PayPal and Reimagine Main Street, May 2025 (a survey of 1,000 small businesses). Owners aren't skeptical of AI; they're skeptical of unsupported rollouts.

PayPal / Reimagine Main Street, May 2025

A number about your business

Which side of the gap is your business on?

The operations scorecard turns this page into a self-assessment: what's documented, what's owned, and whether the systems AI depends on actually exist in your operation. Free, no call required, and you keep the read either way.

Evidence, labeled honestly

Why AI pilots stall, and what changes the odds

The most useful failure data comes from the enterprise world, so I label it that way. MIT's Project NANDA (preliminary findings, July 2025) reported that about 95 percent of enterprise generative AI initiatives showed zero profit-and-loss return on an estimated 30 to 40 billion dollars invested, and that builds done with external partners reached deployment about 67 percent of the time, versus about 33 percent for internal-only efforts. That is an enterprise finding, not a small business one, and it does not mean AI does not work. It means unowned pilots stall, and an outside implementation partner roughly doubles the odds that a build actually ships.

~95% of enterprise GenAI pilots showed zero P&L return

Across an estimated 30 to 40 billion dollars of enterprise spend, roughly 95 percent of generative AI initiatives returned nothing to the P&L. The pattern behind the number: pilots without documented workflows and a clear owner stall before they ship.

MIT Project NANDA, preliminary v0.1, July 2025
External partners deployed about twice as often

In the same research, projects built with external partners reached deployment about 67 percent of the time versus about 33 percent for internal builds. Implementation is a skill. Buying it raises the odds.

MIT Project NANDA, July 2025

The through line

What these numbers add up to

Read together, the numbers say the adoption question is settled and the integration question is not. Most small businesses already pay for AI tools; few have wired them into the workflows the team actually runs, and about half have invested nothing in the training or integration that would change that, per the SBA Office of Advocacy, 2025. That's how a business ends up with a tool graveyard: subscriptions nobody opens, a chatbot nobody trusts, and the same manual process running underneath it all.

That distance is what our AI integration work exists to close, and I'll be plain about what it is and isn't. It isn't more tools, and it isn't a transformation deck. We document the target workflow first, automate the parts that earn it inside the systems you already use, and train your team to own the result, so the engagement ends on purpose. The numbers above name the gap; the method is just closing it deliberately instead of by subscription. If you'd rather check readiness before spending a dollar, the free operations scorecard covers the fundamentals AI depends on.

Straight answers

Questions about the numbers

How many small businesses use AI in 2026?
It depends on the survey and the definition, so I list all three of the credible ones. 76% of small businesses use AI in some form, per Goldman Sachs' 10,000 Small Businesses Voices survey, 2026. 68% use AI regularly, per a QuickBooks-commissioned survey, April 2025. And 58% use generative AI specifically, per the US Chamber of Commerce and Teneo, August 2025. Different questions, different samples, one direction: adoption is mainstream.
What is the small-business AI adoption gap?
The gap is between using AI and running on it. 76% of small businesses use AI, but only 14% have it fully embedded in core operations, per Goldman Sachs, 2026, and about half of AI-using small firms have invested nothing in training or integration, per the SBA Office of Advocacy, 2025. Tools got adopted; workflows mostly didn't change. That distance is an integration problem, not a technology one.
Does AI actually pay off for small businesses?
The self-reported numbers say yes when it gets used: 93% of small-business AI users report a positive impact, per Goldman Sachs, 2026, and 74% of AI users say it makes them more productive, up from 46%, per a QuickBooks-commissioned survey, April 2025. In the US Chamber of Commerce and Teneo study, August 2025, 82% of AI-using small businesses grew their workforce. Self-reported impact is not the same as measured ROI, which is why 74% of owners want clearer ROI evidence, per PayPal and Reimagine Main Street, May 2025. I label the difference instead of rounding sentiment up to proof.
Why do so many AI projects fail or stall?
The best-documented evidence is from enterprises: MIT's Project NANDA, preliminary findings from July 2025, reported about 95 percent of enterprise generative AI pilots produced no P&L return, while builds done with external partners deployed about twice as often as internal ones. The lesson is not that AI does not work. It is that pilots without documented workflows, a clear owner, and implementation experience stall before they ship.
Where do these statistics come from, and can I cite this page?
Every number links to its primary source: Goldman Sachs, Intuit QuickBooks, the US Chamber of Commerce with Teneo, the SBA Office of Advocacy, PayPal with Reimagine Main Street, and MIT's Project NANDA. You're welcome to cite this page, but cite the primaries where you can; that's what they're linked for. When a source revises a figure, this page follows it, and the date at the top changes with it.

Close the gap

Wire AI into the workflows you already run.

If these numbers describe your business, the fix is not another subscription. It's integration: the workflow documented, the automation built into the systems you already use, and your team trained to own it after the engagement ends. That's the whole service, scoped with a defined exit from day one.