US Construction Backlog Dropped in October 2026 Driven by Non-Data Center Sector
Project management · AEC industry
According to Associated Builders and Contractors, the US construction backlog fell to 8.1 months in October 2026. ABC Chief Economist Anirban Basu reported that the decline was primarily driven by worsening conditions for the 88% of contractors not working on data centers, whereas data center contractors expanded their backlog to 11.5 months.
The continuous growth streak in US construction backlog halted in October 2026, according to an Associated Builders and Contractors (ABC) survey conducted between Oct. 20 and Nov. 3. Total backlog dropped month over month to 8.1 months, marking a 0.6-month decline from September and a 0.3-month decrease from the previous year. ABC Chief Economist Anirban Basu attributed the overall drop to deteriorating prospects for the 88% of contractors lacking data center contracts. While contractors engaged in data center construction saw their backlog increase from 11.3 months in September to 11.5 months in October, contractors outside that segment experienced a drop from 8.3 months to 7.6 months. Bookings fell across multiple major categories, including commercial and institutional projects dropping to 8 months and infrastructure ticking down to 8.7 months. Despite lower sales and staffing expectations, contractor sentiment regarding profit margins showed slight improvement.
Key takeaways
- Overall US construction backlog declined to 8.1 months in October 2026, dropping 0.6 months from September and 0.3 months year over year.
- ABC reported a widening divergence: data center contractors held 11.5 months of backlog, while firms without data center work fell to 7.6 months.
- Commercial and institutional backlog dropped by one month to 8 months, while infrastructure backlog decreased to 8.7 months.
- Contractors reported weakened expectations for sales and staffing, though outlook on profit margins saw slight monthly gains.
Source:Construction Dive Read the original ↗
Prepared, checked and automatically published by AI; not individually reviewed by a human.