US labor productivity accelerates in second quarter as firms cut costs with AI

American workers got more done with less effort last quarter, and the numbers weren’t even close to what economists predicted.

Nonfarm business sector productivity rose 1.4% on an annualized basis in the second quarter of 2026, according to Bureau of Labor Statistics data published on August 6. Wall Street had penciled in a 0.6% gain.

The numbers tell a clear story

Output climbed 1.7% while hours worked inched up just 0.3%. Manufacturing productivity was even stronger, gaining 1.9% in the quarter.

Year-over-year, productivity advanced 2.2%. The Q2 number marks a sharp acceleration from the prior quarter’s revised gain of just 0.3%.

Unit labor costs rose 1.3% in the quarter. Real hourly compensation actually declined 3.1%, meaning workers’ purchasing power eroded even as they produced more.

AI is doing the heavy lifting

Five of the largest US banks, specifically Bank of America, Wells Fargo, Citigroup, Goldman Sachs, and Morgan Stanley, collectively cut more than 10,000 jobs in the second quarter. The reductions were directly tied to efficiency improvements and AI integration, according to industry reports.

Coinbase reported notable productivity improvements in engineering roles driven by AI adoption on July 31, just days before the BLS data dropped.

What this means for the Fed and risk assets

The unit labor cost increase of 1.3% is elevated but manageable. Analysts at BNP Paribas have pointed to the sustained productivity gains since 2023 as evidence of a structural shift, anticipating continued improvements through 2026 and into 2027.

The declining real wages component adds another wrinkle worth watching. Workers losing purchasing power while corporate productivity soars could accelerate interest in alternative financial systems and decentralized earning mechanisms.

Productivity gains built on job cuts have a ceiling. The 10,000 banking jobs lost last quarter represent real demand destruction in consumer spending, which could eventually drag on the very output metrics driving the productivity numbers higher.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.