AI Policy Wiki
Dashboard
⚲
AI Policy Wiki › Industries

Construction — AI Deployment

medium confidence · updated 2026-09-17

AI in US construction: agentic project management, jobsite documentation, estimating, safety monitoring, design, and autonomous equipment. AGC/Sage 2026 survey: 61% of firms use AI or plan to increase AI investment (44% a year earlier). The sector is also a supplier to the AI build-out, with data-center projects adding to a construction labor shortage.

Contents

Construction is an early-adoption sector for AI, with applications concentrated in project management, jobsite documentation, safety monitoring, design optimization, cost estimation and, more recently, autonomous earth-moving equipment. It relates to AI policy in two ways. It deploys AI tools amid a persistent labor shortage, and it builds the data centers and power infrastructure that AI companies require, which has made construction employment a recurring argument in data-center siting disputes.

Market and labor context

The US construction market exceeds $2 trillion. The Census Bureau estimated construction spending in July 2026 at a seasonally adjusted annual rate of $2,157.6 billion, of which $1,614.2 billion was private and $543.4 billion public (Source: census.gov). The sector is labor-intensive with low digitization.

Contractors report persistent difficulty hiring. In the 2026 Construction Hiring and Business Outlook from the Associated General Contractors of America (AGC) and Sage, 82% of firms reported difficulty filling hourly craft positions and 80% salaried openings, the highest shares in three years, and 53% expected hiring craft workers to stay as difficult or become harder (Source: agc.org). Associated Builders and Contractors (ABC) estimated that the industry must attract 349,000 net new workers in 2026. It named an aging workforce, immigration enforcement and "rapidly evolving technologies and innovation" among the pressures, and said shortages are more severe around industrial megaprojects such as semiconductor fabrication plants (Source: abc.org).

Adoption patterns

The sector's fragmented contractor structure and project-by-project work create adoption friction. This mirrors, at sector level, the argument in Tyler Cowen's slow-takeoff essay that complementary inputs, integration friction and the weakest link in a production chain slow how quickly AI capability becomes measured output.

Survey evidence shows adoption rising. In the AGC/Sage 2026 outlook, 61% of respondents said their firms use AI or plan to increase investment in it, up from 44% a year earlier. By function, 45% of firms use AI for office and administrative work, 23% for estimating, 20% for design or preconstruction, and 16% for recruitment, training or other HR functions (Source: agc.org). Houzz's 2026 State of AI in Construction & Design report found that more than half of surveyed construction businesses use AI for everyday tasks, and 80% of AI-using firms use it daily (Source: roofingcontractor.com). A software vendor citing International Data Corp. put the share of construction firms using some form of AI at about 60% (Source: openspace.ai). The surveys use different samples and definitions of "use", and their figures are not directly comparable.

The Wall Street Journal reported in late January 2026, under the headline "Construction Companies See Promise in AI Agents", that construction firms were beginning to deploy agentic AI for project coordination, document management and scheduling. The article's subheading describes new AI tools helping managers "stay on top of the complex processes underpinning an exploding sector" (Source: wsj.com). According to OpenSpace, one of the companies featured, reporter Belle Lin framed agents as a response to the labor shortage (Source: openspace.ai; the WSJ article body was not read).

Representative deployments

Reported and emerging applications include:

  • Project management and scheduling. AI agents manage multi-contractor timelines. Procore, a construction-management software provider, announced AI agents in November 2024 to automate routine project tasks, with availability the following year (Source: procore.com).
  • Jobsite documentation and progress monitoring. OpenSpace captures and indexes site imagery so that agents can monitor progress and flag issues (Source: openspace.ai). Erik Brynjolfsson and Andrew McAfee cited site progress checks, which they describe as dull, dirty, sometimes dangerous and expensive, as work suited to automation. Their example is the 2015 partnership between Komatsu and the drone startup Skycatch, in which drones map a site in three dimensions, software compares the map with plans, and the results direct autonomous earth-moving equipment (Machine, Platform, Crowd — McAfee and Brynjolfsson (2017)).
  • Safety monitoring. Computer vision detects hazards on worksites.
  • Design optimization. Generative design is applied to structural and architectural work.
  • Cost estimation. AI assists with quantity takeoffs and bid preparation; 23% of AGC/Sage respondents use AI for estimating (Source: agc.org).
  • Autonomous equipment. Bedrock Robotics, which retrofits excavators, bulldozers and loaders for autonomous operation, raised $270 million in a Series B announced February 4, 2026, co-led by CapitalG and the Valor Atreides AI Fund with participation from Nvidia's NVentures, bringing its total funding above $350 million. It emerged from stealth in July 2025 and completed a supervised autonomy deployment for mass excavation on a 130-acre manufacturing site in November 2025. CapitalG's Derek Zanutto said "the workforce simply isn't there to meet the moment" (Source: prnewswire.com). John Hanke, chief executive of Niantic Spatial, told MIT Technology Review that robots sharing spaces such as construction sites and sidewalks with people will need a human-like level of spatial understanding (Source: technologyreview.com). The AI 2027 scenario projects that most new robots in its fictional timeline are built for factories and construction sites (AI 2027).

Construction as supplier to the AI build-out

The expansion of AI data centers and their power supply is a source of construction demand. An ITIF commentary in January 2026 attributed a 439,000-worker construction shortage to data-center growth. It said data-center construction jobs pay up to 30% more than typical construction jobs, and that firms taking on such projects face backlogs of close to a year (Source: itif.org).

Developers cite construction jobs in project announcements. OpenAI said its SB Energy-built campus at the PORTS Pike site in Ohio would create 35,000 construction jobs and 2,500 permanent operating positions (Source: axios.com; see OpenAI, SB Energy). Meta said its El Paso campus, developed in a venture with BlackRock, would support more than 4,000 construction jobs at peak and 300 operational jobs (Source: investor.atmeta.com; see Meta AI). Research and polling have questioned how much weight these figures carry. Ball State economist Michael Hicks, analyzing data-center openings across 254 Texas counties, found effectively zero net long-term job creation and described construction crews as present "for three weeks" before leaving (Data centers are coming for rural America — Abigail Bassett (The Verge, May 13 2026)). In Public First and ITIF polling, construction jobs moved respondents' choices between data-center proposals by only 2.1 points, the smallest effect of any attribute tested (Public First / ITIF, \"The opposition to data centers\" (polling deck, August 2026); see Data Center Siting / AI Power Politics). Building trades groups have engaged in siting policy: the Pennsylvania State Building and Construction Trades Council said Governor Shapiro's data-center principles offer "a path to build here in Pennsylvania" (Pennsylvania Executive Order 2026-05 (Data Centers)).

Federal and industry workforce initiatives target the trades. America's AI Action Plan (July 2025) calls for training "the workforce that will build, operate, and maintain" AI infrastructure, "including roles such as electricians, advanced HVAC technicians, and a host of other high-paying occupations" (America's AI Action Plan). The White House science strategy of July 2026 similarly argues that "we will still need human hands to build the instrument or prototype" and calls for investment in the skilled trades (Science: A New Golden Age — A Report to the President (OSTP, July 2026)). Mark Zuckerberg wrote in August 2026 that there is a shortage of carpenters, electricians and construction workers to support infrastructure build-out, and that Meta created America's Workforce Academy to train tradespeople with guaranteed jobs near its data centers (The Future is for Everyone (Zuckerberg, August 2026)).

Workforce effects and debates

Economist David Autor uses construction to illustrate how automation affects workers: a construction worker "who is expert with a shovel but cannot drive an excavator will generally experience falling wages as automation advances". He adds that wage gains for workers whose tasks automation complements can be tempered if those skills are abundant elsewhere in the economy (Why Are There Still So Many Jobs? The History and Future of Workplace Automation (Autor, 2015)). Luiza Jarovsky describes an occupational divide in which blue-collar workers in the skilled trades are told their jobs are safe while white-collar workers are told to become AI-fluent. She cites the 2025 inversion of the unemployment gap between bachelor's-degree holders and occupational-associate-degree workers such as plumbers and electricians (The AI Divides Are Here (Jarovsky, April 2026)). Federal Reserve Governor Michael Barr's May 2025 speech set out slow-diffusion and rapid-transformation scenarios for AI's labor-market effects without endorsing either (Fed Gov. Barr — AI and the Labor Market Speech (May 2025)). See AI Labor Disruption.

Relationships

Categories: Industries | us | industry | construction | deployers | ai-agents | project-management | workforce | energy
What links here (7)