Official U.S. statistics do not show AI lifting output in every industry. The Bureau of Labor Statistics (BLS) reported that productivity in the nonfarm business sector grew 2.1 percent in 2025, yet at the industry level the picture was uneven: 41 of 85 manufacturing and mining industries improved, 15 of 30 selected service industries, and 35 of 46 trade industries. None of those releases attributes any change to AI.
Key takeaways
- BLS measures labor productivity as output per hour worked, and it does not separate the part caused by AI from other causes.
- In 2025, productivity rose in 41 of 85 manufacturing and mining industries, 15 of 30 selected service-providing industries, and 35 of 46 trade industries.
- Software publishers posted the largest service-sector gain, 12.9 percent, while wired telecommunications carriers fell 10.8 percent.
- A March 2026 Federal Reserve Bank of Atlanta survey of nearly 750 executives found that perceived AI gains exceed measured ones.
- The 2025 annual industry data are the latest available as of October 2026; BLS has published only quarterly economy-wide figures for 2026 so far.
What the headline productivity number covers
Labor productivity is real output divided by hours worked by all workers, including employees, proprietors, and unpaid family workers. In its revised Productivity and Costs release of March 24, 2026, BLS reported that the nonfarm business sector grew 2.1 percent in 2025, with output up 2.5 percent and hours worked up 0.4 percent. Manufacturing productivity rose 1.9 percent. The release does not mention AI.
That silence is informative. Productivity is a ratio, and it moves with capital investment, staffing choices, demand swings, measurement revisions, and the mix of products. A faster ratio can come from output growing while hours fall, which fits the AI story, but it can also come from a company laying off staff after a slow year. The ratio alone cannot tell the two apart.
What 2025 looked like by industry
BLS also publishes productivity for detailed industries. Its 2025 releases, dated May 28, June 24, and August 26, 2026, show a mixed picture, not a uniform rise.
| Group | Industries with higher productivity in 2025 | Notable movers |
|---|---|---|
| Manufacturing and mining (release of June 24, 2026) | 41 of 85 (39 of 80 in manufacturing, 2 of 5 in mining) | Other transportation equipment manufacturing rose 14.9 percent |
| Wholesale and retail trade (release of May 28, 2026) | 35 of 46 | Wholesale productivity rose 4.4 percent and retail 2.9 percent, with hours falling in both |
| Selected service-providing industries (release of August 26, 2026) | 15 of 30 | Software publishers rose 12.9 percent; wired telecommunications carriers fell 10.8 percent |
In the services release, the three largest gains were software publishers (12.9 percent), line-haul railroads (11.3 percent), and wireless telecommunications carriers (8.8 percent). In each case output rose while hours fell. Couriers and messengers fell 8.5 percent. The manufacturing release also notes that seven industries lost their measures beginning with that release because of changes in source data, so the 85-industry count does not include every industry that existed before.
The trade release stands out for its breadth. Wholesale output grew 3.1 percent while hours fell 1.2 percent, and retail output grew 2.5 percent while hours fell 0.4 percent. The release does not mention AI, and a pattern of more output from fewer hours fits many explanations, from inventory systems to a slow year for hiring.
What 2026 data exist so far
Annual industry figures for 2026 will not appear until mid-2027, judging by the 2025 schedule above. The only 2026 productivity data available in October 2026 are quarterly and economy-wide. In its release of September 3, 2026, BLS reported that nonfarm business labor productivity rose 1.4 percent at an annualized rate in the second quarter of 2026 and 2.2 percent from the same quarter a year earlier, with output up 1.7 percent and hours up 0.3 percent. Manufacturing productivity rose 2.4 percent in the quarter. As in the annual releases, BLS draws no link to AI, and a single quarter is too short to say much about any technology.
Why a gain in one industry is not evidence of AI
Software publishers lead the services list, yet BLS does not say why productivity rose there. A railroad appears just behind it, in an industry with no obvious link to chatbots. The pattern shows how many forces move these numbers at once.
Checking a claim of AI-driven gains in an industry takes more than a single year of data. Readers can ask four things.
- Did output rise, hours fall, or both, and for how many years?
- Did the change begin before widespread generative AI tools were available?
- Is there a firm-level or experimental study that links the change to a specific tool?
- Do other industries with similar conditions show the same movement?
Without answers to these, a productivity gain is a fact about the industry, not about AI.
What executives report versus what statistics measure
Federal Reserve Bank of Atlanta Working Paper 2026-04, "Artificial Intelligence, Productivity, and the Workforce: Evidence from Corporate Executives," was released March 25, 2026. It draws on a survey of nearly 750 executives. The authors report that more than half of firms have invested in AI, that labor productivity gains are positive and vary by sector, and that the largest gains are in high-skill services and finance. They also find that executives perceive larger gains than are measured, which they attribute likely to delayed revenue realization, and that there is little evidence of near-term aggregate job losses.
A survey of managers records beliefs and plans, and those can run ahead of results. The BLS series record what happened to output and hours. The two sources answer different questions, and neither one alone shows that AI is changing every industry.
Limits of this comparison
The BLS industry releases cover only the industries BLS measures, and the services release is a set of 30 selected industries. A count of industries that improved treats a gain of one percent the same as a gain of 15 percent. The 2025 figures are annual, so effects that appear later would not show. The Atlanta Fed survey describes a sample of executives and not the full population of U.S. firms. All of these figures were checked against the published releases in October 2026 and may be revised.
FAQ
Does BLS measure the effect of AI on productivity?
No. BLS publishes output per hour and related measures. Its 2025 releases reviewed for this article do not mention AI or attribute changes to any technology.
Which industries gained the most productivity in 2025?
Among the selected service-providing industries, software publishers gained 12.9 percent, line-haul railroads 11.3 percent, and wireless telecommunications carriers 8.8 percent. In manufacturing, other transportation equipment manufacturing led with 14.9 percent.
Did productivity rise in every U.S. industry in 2025?
No. It rose in 41 of 85 manufacturing and mining industries, 15 of 30 selected service industries, and 35 of 46 trade industries, as published by BLS between May and August 2026.
Why do executive surveys and official statistics disagree about AI?
Surveys capture expectations and perceived gains, while official data capture realized output and hours. The Federal Reserve Bank of Atlanta paper reports that perceived gains exceed measured gains and links the gap to delayed revenue realization.
Related reading
- Reported benefits of AI for industry
- Industries least likely to be displaced
- AI in maritime law
- How AI apps are described and judged
More in AI Tools and Industry.
Last verified: October 2026
Sources
- Productivity and Costs, Fourth Quarter and Annual Averages 2025, Revised, U.S. Bureau of Labor Statistics, March 24, 2026
- Productivity and Costs by Industry: Manufacturing and Mining Industries, 2025, U.S. Bureau of Labor Statistics, June 24, 2026
- Productivity and Costs by Industry: Selected Service-Providing Industries, 2025, U.S. Bureau of Labor Statistics, August 26, 2026
- Productivity and Costs by Industry: Wholesale Trade and Retail Trade Industries, 2025, U.S. Bureau of Labor Statistics, May 28, 2026
- Productivity and Costs, Second Quarter 2026, Revised, U.S. Bureau of Labor Statistics, September 3, 2026
- Artificial Intelligence, Productivity, and the Workforce: Evidence from Corporate Executives, Working Paper 2026-04, Federal Reserve Bank of Atlanta, March 25, 2026