Goldman Sachs figures relayed on X by @pequityresearch at 14:00 UTC on 3 November say AI adoption by firms rose by 0.8 percentage points to 24.6% among US establishments, with adoption expected to rise to 28.6% in the next six months. Both numbers match the US Census Bureau’s Business Trends and Outlook Survey (BTOS), published on 22 October: 24.6% of businesses said they had used AI in the last two weeks in the wave collected from 5 to 18 October, up from 23.8%, and 28.6% expect to use it in the next six months.
That 24.6% is the highest reading since Census began asking the current question in November 2025. Six months before it, in the wave collected from 6 to 19 April, 23.0% of businesses said they expected to be using AI within six months. Actual use came in 1.6 points above that forecast, by our arithmetic, and has been above the earlier forecast in each of the last five readings.
The business lever is demand: how many firms are paying for and using AI tools, which is the market every AI vendor sells into.
Business Pill · Distribution Advantage
An evergreen Business Engineer explainer on distribution advantage: reach decides who gets tried. It teaches the general idea only and does not discuss the Census survey or Goldman Sachs.
The key insight: As we read it, the new element in the Census series is not the level but the direction of the forecast error: for the last five readings, more businesses were using AI than had said six months earlier that they expected to, after six readings in which use fell short of the earlier expectation.
Census asks businesses: “In the last two weeks, did this business use Super Intelligence or SI (formerly referred to as Artificial Intelligence or AI) in any of its business functions?” Its examples include SI-powered data analysis, generative SI for marketing, productdesign or coding, and virtual SI agents for customer service. The Census dashboard labels the series “Current AI Use (Last Two Weeks)”.
In the latest wave, 24.6% answered yes, 65.8% no and 9.6% did not know. Census gives a standard error of 0.27 points on the yes share. The previous two waves both read 23.8%.
The series starts at 17.3% in the wave collected from 17 to 30 November 2025. That is a rise of 7.3 points in under a year, by our arithmetic. A second question asks whether the business will be using it “During the next six months”; 28.6% said yes, 45.5% no and 25.8% did not know.
The gap between expected and current use is 4.0 points, by our arithmetic. In the first wave of the series it was 3.8 points, so expectations have risen roughly in step with use.
Share of US businesses that used AI in the last two weeks (teal) and, plotted 26 weeks later, the share that expected to use it in the next six months (ochre). US Census Bureau BTOS, waves of 17-30 Nov 2025 to 5-18 Oct 2026. Our arithmetic for the alignment.Animated: share of US businesses that used AI in the last two weeks, by Census BTOS survey wave, 17-30 November 2025 to 5-18 October 2026, as published by the Census Bureau. The 7.3-point rise is our arithmetic.
What Was Posted on X
Goldman Sachs: AI adoption by firms rose by 0.8pp to 24.6% among US establishments (with adoption expected to rise to 28.6% in the next 6 months).
$META $GOOGL $AMZN $MSFT $ANT
— P Equity Research ๐ฐ (@pequityresearch) Nov 03 14:00
@pequityresearch posted these Goldman Sachs figures at 14:00 UTC on 3 November with a chart titled ‘Exhibit 8’, whose source line reads ‘Census Bureau, Goldman Sachs Global Investment Research’ and whose footnote says new question wording was used starting 12/4/2025. The 24.6% and 28.6% rates match the Census Business Trends and Outlook Survey estimates published on 22 October, which this piece uses as its primary source.
Firms Keep Underestimating Their Own Uptake
Our check sets each wave’s six-month expectation against actual use 13 waves, or 26 weeks, later. It compares two survey estimates; it does not follow individual firms’ answers.
In the first six pairings, use fell short of the earlier forecast by 0.3 to 0.7 points. Forecasts made in the waves of 17 to 30 November 2025 (21.1%) through 26 January to 8 February 2026 (22.1%) all ran above the use measured half a year later, by our arithmetic.
Then it flipped. The forecast from 9 to 22 February (21.0%) was beaten by 1.4 points in August. The next four were beaten by 1.4, 1.5, 1.3 and 1.6 points, ending with April’s 23.0% against October’s 24.6%.
Where the Use Is: Sectors and Firm Size
The Census dashboard breaks use out for six sectors. In the 5 to 18 October wave, Information read 46.4%, Professional, Scientific, and Technical Services 43.8%, Finance and Insurance 40.5%, Educational Services 33.4%, Manufacturing 22.9% and Retail Trade 18.5%.
Manufacturing was 12.7% in the November 2025 wave and Finance and Insurance 29.9%. Sector readings move more from wave to wave: Information was 50.2% and Educational Services 39.9% in the wave before.
By size, 43.6% of businesses with 250 or more employees used AI, up from 30.5% in November 2025. Among businesses with one to four employees it was 24.3%, up from 17.0%. The gap between the two widened from 13.5 to 19.3 points, by our arithmetic, though the 250-plus reading was 36.5% one wave earlier.
How Goldman’s Number Lines Up
The relayed figures, 24.6%, 28.6% and the 0.8-point rise, are the same as the Census estimates to one decimal place. The post did not link a Goldman Sachs document, and we have not read one.
The relay says “among US establishments”; Census describes the estimates as businesses. Census published the wave on 22 October, 12 days before the relay. Its next release, for the wave collected from 19 October to 1 November, is scheduled for 5 November.
An older Census question asked whether a business used AI “in producing goods or services”. It read 10.0% in its last wave, collected from 22 September to 5 October 2025. That is a narrower question and is not comparable with the current series. Census collected no data for 6 October to 16 November 2025 because of the federal funding lapse.
As we read the figures, the relayed Goldman Sachs numbers add no new measurement: 24.6%, 28.6% and the 0.8-point rise are the Census estimates published on 22 October, to one decimal place.
As we read the size breakdown, the largest businesses moved most: the 250-plus class rose 13.1 points from November 2025 to 43.6%, against 7.3 points for businesses with one to four employees, by our arithmetic; single-wave readings in the size classes move more than the national figure.
As we read the expectations question, about a quarter of businesses (25.8%) still answer that they do not know whether they will use AI in the next six months, a larger share than the 9.6% who do not know whether they used it in the last two weeks.
US Census Bureau, Business Trends and Outlook Survey, question 24
“During the next six months, do you think this business will be using Super Intelligence or SI (formerly referred to as Artificial Intelligence or AI) in any of its business functions?”
Three Implications
FOR AI VENDORS The Census estimate counts businesses that used AI in the last two weeks: 24.6% in the 5 to 18 October wave, up from 17.3% in November 2025.
FOR SECTOR COMPARISONS Information (46.4%), Professional, Scientific, and Technical Services (43.8%) and Finance and Insurance (40.5%) lead the six sectors on the Census dashboard; Retail Trade reads 18.5%.
FOR THE NEXT RELEASE Census schedules the wave collected from 19 October to 1 November for publication on 5 November, which adds one more pairing to the forecast check.
The framework argues: “Model capability is exponential; product capability is approximately linear.”
As we read it, the Census forecast gap is one measurable trace of that kind of gap on the business side: firms planning on a steady path have, for five readings, under-called their own adoption. The framework does not discuss the Census survey or business adoption rates.
What Is Not Established
It is not established whether Goldman Sachs adjusts the Census data, for example by weighting by employment, beyond the two figures in the relay. We have not seen Goldman’s note.
The two questions ask whether a business used AI, not how much it used or what it spent. The current question text refers to Super Intelligence or SI, “formerly referred to as Artificial Intelligence or AI”; the files we read do not say from which wave respondents saw that wording. Whether use stays above the earlier forecasts is not known; the 5 November release adds one more pairing.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
The 24.6% Goldman figure relayed on X is the Census BTOS estimate for 5 to 18 October, published on 22 October, and the highest since the current question began. By our arithmetic, use is 1.6 points above the 23.0% firms forecast in April, the fifth reading in a row above the earlier forecast.
A note on sourcing. This piece is based on the US Census Bureau’s Business Trends and Outlook Survey files (National.xlsx and the dashboard’s AI sector and employment-size files), downloaded on 3 November 2026, and a post on X relaying Goldman Sachs figures. We have not read a Goldman Sachs report. The record, the 1.6-point forecast gap and the other differences are our arithmetic from the Census estimates. We did not contact Goldman Sachs or the Census Bureau.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
