🔍 Read the full analysis: Why Factory Work Is Retreating As AI Advances on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
U.S. factory employment reached a record 19.6 million in June 1979 and stood at a preliminary 12.7 million in October 2026, according to the source’s employment data. Factory output was nearly twice its 1979 level in September 2026, though still about 9% below its 2007 peak. The figures show a long decline in factory jobs, but they do not establish that AI caused it.
U.S. factories employed a preliminary 12.7 million people in October 2026, about 35% below the record set in June 1979, while factory output in September was nearly twice its 1979 level, according to a report published November 8 by ThorstenMeyerAI.com. The comparison captures the scale of manufacturing’s employment retreat, but the data cited do not establish that artificial intelligence caused the decline.
The report uses the U.S. Bureau of Labor Statistics’ Current Employment Statistics series, which counts jobs on employer payrolls outside farming. It excludes self-employed workers and people employed in private households. The series begins in 1939. Factory employment reached 19.553 million in June 1979, its highest reported level, then fell to 12.7 million in October 2026, a preliminary figure. Manufacturing accounted for 8.0% of payroll jobs in October, compared with 21.7% in June 1979.
The employment share reached its high much earlier: factories represented 38.8% of U.S. payroll jobs in November 1943, during World War II. The number of factory jobs continued to rise for decades after that share began falling, as the overall economy expanded. The report says the jobs total fell particularly sharply between January 2000 and February 2010, from 17.3 million to 11.4 million—almost six million jobs.
Output followed a different course over the long term. The Federal Reserve’s manufacturing production index rose from 50.4 in June 1979 to 98.4 in September 2026, using a 2017 base of 100. The report calculates that output per factory job was roughly three times its 1979 level, while cautioning that this rough comparison counts jobs, not hours worked. Output has not risen continuously: it peaked in December 2007, and the September 2026 reading was about 9% below that peak.
Fewer Jobs, More Output
The figures show why factory employment and industrial production should not be treated as the same measure. Over the period from 1979 to 2026, employment fell while measured output rose; that does not mean manufacturing stopped mattering to the economy, nor does it identify which specific technologies or business changes explain the difference. The decline in factory jobs matters to workers and communities that depend on manufacturing payrolls, while the output figures show that the sector can produce more over time with fewer recorded jobs.
The distinction is also relevant to claims about AI and work. A long-running decline that began decades before today’s generative AI systems cannot, from these totals alone, be attributed to AI. Automation and other productivity changes may be part of the broader story, but the source does not provide a measure separating AI’s effects from other forces. Its own output series also complicates a simple account of relentless growth: production remains below its 2007 high.
Products Worth Considering
Sony WH-1000XM4 Wireless Noise-Canceling Headphones, Silver
Apple MacBook Air 13.6-inch Laptop with M5 Chip, 24GB Memory, 1TB SSD, Midnight
Amazon Kindle Paperwhite (2024, 12th Generation), 16 GB, 7-Inch Display, Black
As an affiliate, we earn on qualifying purchases.
A Long Shift in Factory Work
The U.S. pattern developed over many decades. In January 1939, the BLS series recorded about 9.1 million factory payroll jobs, equal to 30.3% of all payroll employment. Wartime production brought the share to its 1943 peak. After the war, the share gradually declined even as the number of factory jobs continued to rise, reaching its record in 1979.
The source places the later U.S. decline alongside changes in Germany and China, but the available figures are not directly comparable because the countries’ measures cover different sectors and periods. In Germany, industry and construction accounted for 22.9% of employment in 2025, compared with 49.2% for the broader producing sector in West Germany in 1965; the source notes changes in classification. It also reports that China’s industry and construction employment peaked in 2012 at 232 million. These comparisons suggest that shrinking industrial employment shares are not unique to the United States, but they do not show that the same causes operated in each country.
The report also challenges the claim that U.S. factory jobs simply “went to China.” It says the U.S. decline began 22 years before China joined the World Trade Organization. Citing economists’ estimates, it says rising Chinese imports cost up to about one million U.S. factory jobs from 1999 to 2011, compared with almost six million lost across that period. That estimate describes one contributing factor, not a complete accounting of the decline.
“The number to remember is 19.6 million.”
— ThorstenMeyerAI.com report
What the Data Cannot Show
The employment and output series document changes in jobs and production, but they do not measure how much of the employment decline came from AI, other automation, trade, recessions, or shifts in the kinds of goods produced. The source provides no industry-by-industry or task-level estimate of AI’s effect. Its description of factory work retreating “as AI advances” should therefore be read as a framing of a broader question, not evidence that AI drove the historical decline.
The latest employment total is preliminary, and the output data end one month earlier. The report’s “about three times as much output per factory job” calculation compares an output index with job counts; it does not account for hours worked, changes in the mix of products, or differences among manufacturing industries. The available information also does not establish whether output or employment will rise or fall after the latest reported months.
Watch the Next Data Releases
The next U.S. employment and industrial production releases will show whether the October and September readings hold as later data arrive. Revisions could alter the preliminary employment figure. To assess AI’s role more directly, readers would need evidence that links technology adoption to changes in staffing or tasks at particular manufacturers, rather than relying on national employment and output totals alone.
Further comparison will also depend on consistent definitions across countries and time. The source notes that Germany’s sector classifications changed, while the U.S. payroll series and the German and Chinese measures cover different populations or industries. Those limits matter when deciding whether the reported patterns share common causes.
Key Questions
How many people worked in U.S. factories at the 1979 peak?
The BLS payroll series recorded 19.553 million factory jobs in June 1979, commonly rounded to 19.6 million.
How many U.S. factory jobs were reported in 2026?
The report cites a preliminary 12.7 million jobs in October 2026, equal to 8.0% of payroll employment.
Did factory output fall along with employment?
Not across the full period cited. The Federal Reserve manufacturing index was nearly twice its June 1979 level in September 2026, but it remained about 9% below its December 2007 peak.
Does this report show that AI caused factory job losses?
No. The figures show long-term changes in employment and production, but do not isolate AI’s effect from trade, other technologies, economic downturns, or changes in manufacturing.
Did U.S. factory jobs all move to China?
The report says rising Chinese imports may have contributed to job losses: economists estimate up to about one million U.S. factory jobs were lost to that factor from 1999 to 2011, out of almost six million lost overall. That estimate does not account for the full decline.
Source: ThorstenMeyerAI.com
