The AI Gap: How UK Construction Is (and Isn’t) Adopting Artificial Intelligence in 2026

Executive Summary

Artificial intelligence is the most-discussed technology in construction — and the least adopted. Across every credible UK survey published in the last twelve months, construction sits at or near the bottom of the league table for AI adoption, even as investment intent, tooling maturity and proven use cases accelerate around it.

The key findings of this report:

• Construction is the UK’s least AI-adopted major sector. Cross-industry benchmarking puts construction adoption at around 6% — the lowest of any industry surveyed — and government research finds that roughly nine in ten construction businesses neither use AI nor plan to.
• The SME picture is starker still. The Federation of Small Businesses found just 1% AI adoption among small construction firms, against 37% in professional and technical services.
• But adoption figures conceal a paradox. Firms that do adopt AI use it intensively — in micro businesses that adopt, 38% of staff use AI, roughly double the rate inside large adopting firms. When a five-person contractor adopts AI, it becomes everyone’s assistant, not a department.
• The gap between intent and implementation is wide. In RICS’s global survey (nearly half UK respondents), 56% of investors planned to increase AI spending, yet 45% of organisations reported no AI implementation at all and a further 34% were only in early pilots.
• Early adopters are reporting measurable returns. Professionals using AI report saving over three hours a week, AI-assisted cost estimation has been shown to cut budget overruns by 13–20%, and the direction of travel in safety, estimating and site monitoring is unmistakable.

The conclusion: UK construction’s AI problem is not a technology problem. The tools exist, the returns are demonstrable, and the sector’s chronic pressures — thin margins, labour shortages, rework and admin burden — are precisely the problems AI addresses. The barriers are skills, data quality, integration and, above all, the structure of an industry dominated by small firms with no time to experiment.

1. The State of Adoption: Bottom of the Table

What the surveys show

Depending on which survey you read, UK business AI adoption is anywhere from 16% to over 50% — the figures vary with sample, definition and fieldwork date. But every credible source agrees on the ranking, and construction is last:

• The Department for Science, Innovation and Technology’s AI adoption research, published in January 2026, found that around nine in ten construction, hospitality, retail and transport businesses neither use AI nor plan to — while information and communication leads at 43%.
• The ONS finds 25% adoption across all businesses, rising to 44% among firms with 250 or more staff — a size effect that works directly against a sector where the overwhelming majority of firms are micro and small businesses.
• Cross-industry benchmarking of DSIT, ONS and ONS-adjacent data puts construction sector adoption at approximately 6%, the lowest of all industries surveyed, against 43–51% in ICT and 21–31% in financial services.
• The FSB’s small business research found 37% adoption among professional and technical small firms — and 1% in construction.

Within the industry’s own research, the picture is more nuanced but consistent. RICS’s Artificial Intelligence in Construction report (48% of respondents UK-based) found roughly 45% of organisations with no AI implementation and 34% in early pilot phases — cautious experimentation rather than operational use. Meanwhile CITB digital skills data suggests around 23% of construction businesses use some form of AI-powered tool, against 67% using mobile devices on site and 42% using digital project management software.

Why the numbers disagree — and why it matters

The spread between “1% of small construction firms” and “23% use an AI-powered tool” is not a contradiction; it is a definition problem. Ask a contractor whether they have “adopted AI” and most will say no. Ask whether anyone in the office uses ChatGPT to draft a quote, a tender response or a RAMS document, and the answer is very different. A significant amount of AI use in construction is informal, individual and invisible to surveys — which means the real adoption floor is higher than the headline figures suggest, but the strategic adoption rate (AI embedded in how the business actually operates) genuinely is close to the bottom of the table.

The intensity paradox

The most under-reported finding in the national data: adoption and intensity run in opposite directions. DSIT found that in micro firms that have adopted AI, 38% of staff use it — against 26% in small firms, 18% in medium and 20% in large firms. Large companies adopt more often; small companies use AI harder when they do. For SME contractors, this is the key strategic insight — AI does not require a transformation programme, a data team or an IT department. It requires one decision and a few champions.

2. Where AI Is Actually Being Used

Beyond the hype, five use case clusters are delivering measurable value on UK projects today:

Estimating, tendering and admin. AI-assisted cost estimation has been shown to reduce budget overruns by 13–20%. For SMEs, the fastest wins are more mundane: drafting and reviewing tenders, generating quotes from historical data, scanning and coding invoices, producing method statements and RAMS documents, and summarising contract terms. This is where the informal ChatGPT-style adoption already happening in back offices lives — and where hours per week are being recovered right now.

Design and optioneering. In RICS’s survey, design optioneering was identified as the frontier: 40% of respondents expect AI’s biggest impact over the next five years to be in producing smarter, faster project design. Generative design tools can produce and compare design options against cost, carbon and buildability constraints in hours rather than weeks.

Site monitoring and progress tracking. AI-powered analysis of drone imagery, fixed cameras and 360° site capture allows as-built conditions to be compared automatically against the design model, flagging deviations, quantifying progress and feeding valuations. Reality capture data increasingly flows into digital twins — virtual replicas of the project used for progress tracking, logistics planning and quality assurance.

Safety. Computer vision systems monitor sites through existing cameras to detect unsafe behaviours, missing PPE, exclusion zone breaches and plant-person proximity risks as they occur, shifting safety management from reactive investigation to real-time prevention. Sensor data combined with AI analytics can also flag hazardous conditions before incidents happen.

Predictive planning and maintenance. Machine learning models trained on historical project data identify schedule risk points before they materialise, while predictive maintenance on plant and equipment reduces downtime — an approach that carries directly into building operation, where digital twins deliver some of their strongest returns.

What early adopters report

Houzz’s inaugural UK State of AI in Construction and Design report (2026) — a small sample of around 145 professionals, weighted towards design — found 46% already using AI, reporting average time savings of over three hours per week and an estimated £23,000 annual productivity benefit. Treat the specific figures with caution given the sample, but the pattern matches the broader evidence: where AI is adopted for defined tasks, the payback is quick and the users don’t go back.

3. What’s Holding the Industry Back

The barriers are consistent across every survey:

• Skills. Lack of skilled personnel is the most-cited obstacle — 46% in RICS’s survey — and cross-sector UK research similarly identifies the skills gap as the primary barrier to adoption. This compounds construction’s existing workforce crisis: the sector struggling hardest to recruit is also the one least equipped to retrain.
• Data quality and fragmentation. Around 30% cite poor data quality. Construction’s data is scattered across emails, PDFs, spreadsheets, WhatsApp threads and site diaries. AI is only as good as the data it can reach, and most contractors’ data cannot currently be reached.
• Integration. 37% cite the difficulty of connecting AI tools with existing systems and workflows. Point solutions that don’t talk to the accounts package, the programme or the document management system create work rather than removing it.
• Structure and margins. An industry of small firms operating on low single-digit margins has no slack for experimentation, no dedicated IT function, and a justified scepticism of software promises. The DSIT finding that most construction firms have no plans to adopt reflects rational prioritisation under pressure, not ignorance.
• Trust and accountability. Liability in construction is unforgiving, and professionals are rightly cautious about black-box outputs. Governance is beginning to catch up: RICS’s first global professional standard for the responsible use of AI in surveying practice took effect for all members and regulated firms on 9 March 2026 — a signal that AI use in the built environment is moving from informal to professionally accountable.

4. Why This Matters More in Construction Than Anywhere Else

Construction’s low adoption is usually framed as laggardness. The more useful framing is exposure: the sector’s biggest structural problems are the ones AI is best placed to relieve.

• The labour shortage. As covered in our skills gap report, the industry needs an extra 41,200 workers a year to 2030 and cannot recruit them fast enough. Productivity gains are not optional — CITB itself notes that productivity improvements have so far been insufficient to close the workforce gap. Every hour of estimating, admin or rework that AI removes is an hour of scarce labour returned to billable work.
• Margins. With tender price inflation, wage growth above 7% in shortage trades, and higher employment costs from recent Budgets, contractors have almost no pricing headroom. A 13–20% reduction in budget overruns is the difference between profit and loss on many jobs.
• Compliance burden. The Building Safety Act’s golden thread requirements are, at heart, an information management problem — exactly the category of problem AI-assisted documentation, search and audit tools address.

The competitive implication is uncomfortable for late movers: in a sector where 90% of firms have no AI plans, the 10% that build even modest capability gain a durable edge in bid speed, pricing accuracy and overhead — and the intensity paradox shows small firms can capture it fastest.

5. A Practical Adoption Path for SME Contractors

The evidence points to a clear sequence — none of it requiring a transformation budget:

1. Start with documents, not sites. The highest-return, lowest-risk entry point is the back office: tender drafting, quote generation, RAMS and method statements, meeting summaries, contract review. General-purpose AI tools handle these today for tens of pounds per user per month.
2. Fix data habits before buying platforms. Standardise where project information lives and how it’s named. Every future AI use case depends on it, and it costs discipline rather than money.
3. Pilot one site-facing use case with a defined metric. Progress capture, drone surveys or safety monitoring on a single project, with a before/after measure — hours saved, rework avoided, incidents flagged. Scale only what pays.
4. Put a human accountability line around every output. AI drafts; a competent person checks and signs. This is both good governance and, increasingly, professional obligation as standards like the RICS AI standard take hold.
5. Train the willing, not everyone. The intensity data shows adoption spreads through champions. One person per team who uses the tools daily will move the business further than a company-wide training day.

6. Conclusion

UK construction enters the second half of 2026 with the widest gap of any sector between what AI can demonstrably do for it and how much of it is being used. The barriers — skills, data, integration, margin pressure — are real but not fixed, and the firms overcoming them are reporting concrete returns in hours, accuracy and overhead.

The industry does not need to lead the AI race. It needs to stop finishing last in a race whose prizes — productivity, margin resilience and relief from a workforce shortage that cannot otherwise be closed this decade — are worth more to construction than to almost any other sector. The tools are ready. The question for every contractor is no longer whether AI applies to construction, but whether their competitors adopt it first.

Sources

• Department for Science, Innovation and Technology, AI Adoption Research (January 2026)
• Office for National Statistics, business AI adoption data
• Federation of Small Businesses, small business AI adoption research
• RICS, Artificial Intelligence in Construction Report (2025) and RICS professional standard on responsible AI use (effective March 2026)
• Helium42, AI Adoption Benchmark Report 2026 (synthesis of DSIT, ONS, PwC, Deloitte, McKinsey and Stanford AI Index data)
• Houzz, UK State of AI in Construction and Design Report (2026)
• CITB, digital skills and technology adoption data
• Industry analysis of AI use cases in estimating, site monitoring, digital twins and safety (2025–2026)