The median UK contract day rate for roles citing AI is £559, up 1.64% on a year earlier, across 3,018 quoted rates. Data scientist rates rose 12.94% to £600 in the same window. The contract market has stopped paying extra for the word and pays for skills it can test before go-live. The market-size numbers are less settled. Published estimates of the 2025 global AI consulting services market run from $7.39bn to $43.66bn, and two come from one publisher.
These AI consulting statistics each carry a publisher, a sample or a page date, and a link to the page we checked on 28 September 2026. Several make the opportunity look smaller, not larger. They stay in, because a benchmark that reports only the flattering half is a brochure.
AI consulting statistics at a glance
Published figures on the AI consulting market, adoption, buyer attitudes, rates, talent, regulation and insurance, each attributed inline with its sample or page date and linked in the sources below.
- Selection rule
- Survey figures are included only where the source names its sample and reference period. Market-size estimates are models, not surveys, so each is shown with its publisher and page date, side by side.
- Primary sources
- SPI Research 2026 Professional Services Maturity Benchmark (509 organisations), MCA Member Survey 2026 (1,000+ consultants) and Client Survey 2026 (350+ client leaders), Source Global Research (3,887 companies), Pave's 2025 AI & ML report, IT Jobs Watch, six market research reports, the EU AI Act as amended, and insurance analysis from Fenwick and Hunton.
- Snapshot dates
- IT Jobs Watch figures are rolling six-month medians. Contract rates are to 21 August 2026, matching our AI rates benchmark. Permanent salaries are to 28 September 2026.
- Deliberate inclusions
- Figures that weaken the case for AI consulting sit alongside those that strengthen it.
- Review cycle
- Reviewed quarterly. Every figure was checked against its linked source on 28 September 2026.
ConsultingDemand, “AI Consulting Statistics: Market Size, Rates and Trust”, data as of 28 September 2026. https://consultingdemand.com/blog/ai-consulting-statistics/
How big is the AI consulting services market?
Nobody agrees, and the gap is not a rounding error. We found six estimates on the research publishers’ own pages, all in US dollars. For 2025 they run from $7.39bn to $43.66bn.
| Publisher and report | Page date | Market size | Forecast | CAGR |
|---|---|---|---|---|
| The Business Research Company, AI Consulting Market | September 2026 | $7.39bn (2025) | $19.47bn by 2030 | 21.4% |
| SNS Insider, AI Consulting Services Market | 15 May 2026 | $10.86bn (2025) | $94.41bn by 2035 | 24.14% |
| Future Market Insights, AI Consulting Services Market | 12 June 2025 | $11.07bn (2025) | $90.99bn by 2035 | 23.4% |
| TechMarket Experts, AI Consulting and Support Services | 23 September 2025 | $14bn (2024) | $72.8bn by 2030 | 31.6% |
| Market Research Future, AI Consulting Services Market | 30 March 2026 | $36.6bn (2025) | $120bn by 2035 | 12.61% |
| Market Research Future, AI Consulting Service Market | 17 September 2026 | $43.66bn (2025) | $197.18bn by 2035 | 16.27% |
CAGR is the compound annual growth rate each publisher states. SNS Insider’s page prints 24.14% in its summary and 24.41% in its scope table; 24.14% is the rate its own start and end figures produce.
The largest 2025 estimate is nearly six times the smallest. Both ends describe the same year, so the gap is in what each report counts and how, not in the forecast. Market Research Future publishes two reports on what reads as the same market, and they differ by just over $7bn for the same year. TechMarket Experts counts consulting and support services together, a wider scope by its own title.
The forecasts spread further. Stated growth rates run from 12.61% to 31.6% a year, over different periods, so the 2035 figures range from $90.99bn to $197.18bn.
None of these is a survey with a sample. Each is a model sold as a report. If a supplier’s deck cites a market size, ask which report and which year. If it cannot say, the number is decoration.
Adoption
1. 27.1% of professional services projects used generative AI in 2025, up 40% from 19.3% in 2024. (SPI Research 2026 Professional Services Maturity Benchmark, 509 organisations, as published by its sponsor Deltek)
2. 77% of MCA member firms have integrated AI into their systems or enabled employees to use AI models. (MCA Member Survey 2026, over 1,000 consultants, run by Savanta for the Management Consultancies Association, published 20 January 2026)
3. Firms using AI widely in service execution, with measurable benefits, posted 17.9% EBITDA (earnings before interest, tax, depreciation and amortisation) and 81.5% on-time delivery. Firms using it experimentally posted 9.0% and 74.2%. Firms not applying it posted 6.0%. (SPI Research 2026 Benchmark, as reproduced by Rocketlane)
That third figure is the strongest argument in the dataset, and it shows correlation, not cause. Firms that run delivery well adopt tools well, and are profitable for other reasons. The same table carries a warning. Firms calling AI “fully embedded and strategic” posted 6.3% EBITDA, but they are 1.9% of those surveyed, a cohort flagged as too small for conclusive findings. Treat 17.9% as a reason to ask a supplier what “widely used” means in their delivery, not as a margin forecast.
The wider industry figures, from utilisation to UK market size, sit in our consulting industry statistics.
What buyers think
4. 70% of 3,887 companies polled said they would not trust a consultancy report produced with AI. Almost a third said learning a firm had used AI in a published report would shake their confidence in it. (Source Global Research poll, reported by The Telegraph and Resultsense, 4 August 2026)
5. Among companies that had used consultants’ AI tools, 77% thought the technology was a bubble about to burst. (Source Global Research)
6. Among those that had not, the figure was 55%. That is a 22-point gap, and it runs the wrong way for anyone selling the tools. (Source Global Research)
7. 40% of UK consulting clients named data privacy and security as the biggest challenge in adopting AI solutions. (MCA Client Survey 2026, 350+ client leaders, run by Savanta, published 6 May 2026; multiple choice)
8. 34% named a lack of skilled professionals. In the public sector the figure was 45%, against 32% in the private sector. (MCA Client Survey 2026)
9. 29% named high implementation costs. (MCA Client Survey 2026)
10. 28% named difficulty defining a clear AI strategy and roadmap. (MCA Client Survey 2026)
The four barriers the MCA published are security, skills, cost and strategy. None of them is about what the models can do. A supplier pitching model capability is answering a question your board is not asking. Ask instead for the data-protection controls, the skills transfer and the cost of running the thing in production.
The same survey found clients judge a finished project by delivery of agreed outcomes, relevance to their problem and value for money. They also named skills transfer as a critical requirement. The MCA’s release on the Client Survey 2026 carries this line from its chief executive.
The MCA Client Survey 2026 shows the competitive edge now lies in being able to provide judgement, sector knowledge and trust that is earned through outcomes.
Pricing and rates
11. UK contract vacancies citing AI carry a £559 median day rate, up 1.64% from £550. (IT Jobs Watch, 3,018 quoted rates, six months to 21 August 2026)
12. That is the largest sample in our AI consulting rates benchmark, and the flattest rate. AI is a skill marker, not a job title: IT Jobs Watch counts every contract role that cites it, so it does not describe AI consultants specifically. (IT Jobs Watch)
13. Machine learning rates rose 3.42% to £575, from £556. (IT Jobs Watch, 1,018 quoted rates)
14. Data scientist rates rose 12.94% to £600, from £531, roughly eight times the AI rate of increase. (IT Jobs Watch, 217 quoted rates)
15. US federal ceiling rates for AI and data science roles run $138 to $208 an hour for the middle half, median $171, the highest median of any industry on the schedules. A ceiling rate is the most a contractor may charge a US federal agency for that labour category. (Consulting fees by industry, from 1,602 rates in GSA CALC+, 28 September 2026)
16. KPMG International pressed its auditor, Grant Thornton UK, to pass on AI efficiencies. The fee for auditing its 2025 accounts fell 14%, to $357,000 from $416,000. (Financial Times, republished by The Irish Times, 6 February 2026, citing UK filings)
Price the contractor and the consultancy separately. £559 buys a contractor’s day. A firm that carries delivery risk prices it in: on US federal schedules, AI and data science roles carry ceiling rates of $138 to $208 an hour. The efficiency story that wins an AI engagement is also the one your procurement team can use to cut the fee, as KPMG showed. On a fixed-price SoW, ask for the effort estimate behind each workstream.
If you are still deciding whether you need outside help at all, read when to hire a tech consultant before you compare rates.
What the talent costs
17. The median UK permanent salary for jobs citing AI is £74,000, up 5.71% on a year earlier. (IT Jobs Watch, 5,716 salaries quoted, six months to 28 September 2026)
18. In London the median is £85,000. Across the UK excluding London it is £65,000. (IT Jobs Watch, London sample 2,278 salaries, same window)
19. Data engineers, whom most AI programmes need before they need a model, have a £70,000 UK median, unchanged on a year earlier. (IT Jobs Watch, 1,513 salaries quoted, six months to 28 September 2026)
20. Annual attrition for AI and machine learning individual contributors runs at 28%, and 22% for AI/ML managers. (Pave, 2025 AI & ML Compensation Trends & Practices Report, published 11 February 2025)
21. For software engineers the figure is 17% at both levels. (Pave)
An eleven-point attrition gap belongs in the build-versus-buy case. A team losing 28% of its people a year hires continuously to stand still. The same applies to your supplier’s bench, so write knowledge transfer and runbooks into the SoW (statement of work), and the context stays when their people rotate.
The regulatory picture
22. The EU AI Act’s high-risk obligations now apply from 2 December 2027 for the stand-alone uses listed in Annex III, and 2 August 2028 for AI in products covered by Annex I. The Digital Omnibus, Regulation (EU) 2026/1744 of 8 July 2026, moved them. It entered into force on 27 July 2026. (Digital Omnibus on AI, full text on the EU AI Act Explorer)
23. The rest of the Act, including the Article 50 transparency duties, applies from 2 August 2026. Providers of generative systems already on the market before that date have until 2 December 2026 to meet the Article 50(2) marking duty. (EU AI Act implementation timeline)
24. Article 4, as amended by the Omnibus, requires providers and deployers to take measures to support the AI literacy of their staff. It no longer asks for a “sufficient level” of literacy, and it does not require them to guarantee any individual’s. (EU AI Act, Article 4, as amended)
The high-risk rules were originally due on 2 August 2026, the Act's general date of application. The Omnibus moved them six days before that date arrived. Anything telling you high-risk obligations applied from August 2026 was written before 27 July 2026, or did not check.
For a SoW signed now, name who carries the Article 50 disclosure duty for anything the supplier builds, and which literacy measures cover the staff who will run it.
Insurance and liability
25. Several insurers have introduced “absolute” AI exclusions in Tech E&O (errors and omissions) and professional liability policies. Similar wording is appearing in D&O (directors and officers), employment practices and fiduciary policies. (Fenwick, 15 June 2026)
26. Berkley’s version, written for its D&O, E&O and fiduciary liability products, excludes claims relating to “any actual or alleged use, deployment, or development of Artificial Intelligence”. (Hunton Insurance Recovery Blog, 28 May 2025)
27. A professional services firm creates AI exposure when its staff use third-party generative AI tools, even if it sells no AI product. Fenwick warns that policy wording can move that loss outside cover while the policy still looks AI-neutral. (Fenwick)
28. UK professional indemnity (PI) underwriters are beginning to ask how firms use AI and who reviews its output. (Professional Indemnity Insurance Brokers, July 2026 market analysis)
Ask your supplier for written confirmation that its PI or Tech E&O policy covers AI-assisted work, and name the policy in the contract. The rest of the contract, from brief to signature, is in our guide to how to hire a consultant.
Vendor partner programmes
29. Anthropic launched the Claude Partner Network on 12 March 2026 with an initial $100 million for 2026. Membership is free to any organisation bringing Claude to market. (Anthropic)
30. OpenAI launched its Partner Network on 14 June 2026, backed by $150 million, with Select, Advanced and Elite tiers. It aims to train and enable 300,000 certified consultants by the end of 2026. (OpenAI)
Partner status is an on-ramp, not a moat. A certification OpenAI wants 300,000 consultants to hold does not tell you which of them has put a workload into production. Ask for the deployment, not the badge.
What the numbers add up to
Nobody can tell you the size of the AI consulting market to within a factor of five. Adoption is broad, and the margin gain sits with firms using AI widely in delivery. Buyers grow more sceptical the more of the tooling they see. Contract rates for the headline skill are flat while the testable skills beside it rise. And the efficiency story that wins the pitch is the one that cuts the fee.
So put it in the SoW. Five clauses carry most of the weight:
- Acceptance criteria tied to production. A named workload, live, on your data, meeting a stated threshold. A demo does not count.
- AI-use disclosure. Which deliverables were drafted with AI, and who reviewed them. Seven in ten companies in the Source Global Research poll would not trust an AI-produced report.
- Data-protection controls. Where your data goes, which models see it, and what is retained. It is the barrier 40% of MCA clients named.
- Insurance confirmation. Written confirmation that the supplier’s PI or Tech E&O policy covers AI-assisted work, with no absolute AI exclusion.
- Knowledge transfer and runbooks. With AI/ML attrition at 28% a year, the context has to live in documents, not in people.
Add a sixth if the system falls under the EU AI Act: who carries the Article 50 transparency duty, and from which date.
Our guide to hiring an AI consultant has the full question set for the supplier meeting.
If you run a consultancy and sit on the other side of this table, our guide to whether your consulting firm should offer AI services works through the build-versus-partner economics.
- Published estimates of the 2025 AI consulting services market run from $7.39bn to $43.66bn, and one publisher issued two different figures
- 27.1% of professional services projects used generative AI in 2025, up from 19.3%, and 77% of MCA member firms have AI in their systems
- Firms using AI widely in delivery posted 17.9% EBITDA against 9.0% for experimenters, a correlation rather than demonstrated cause
- 70% of 3,887 companies would not trust an AI-produced report, and those who had used consultants' AI tools were more sceptical, not less
- UK contract rates citing AI rose 1.64% to £559 on 3,018 rates, while data scientist rates rose 12.94% to £600
- EU AI Act high-risk duties now start on 2 December 2027 and 2 August 2028, while Article 50 transparency duties apply from 2 August 2026
Last updated: 28 September 2026