AI in Accounting: 58 Statistics for 2026 (With Sources and Dates)

Most statistics pages in this category repeat each other's numbers without saying where they came from or what year they describe. Several of the most-quoted figures contradict each other outright.

This page does three things differently. Every figure carries the period it covers and its unit. Every figure names its source. And where credible sources disagree, we show both numbers rather than picking the flattering one.

Last reviewed: September 2026.

---

Market size

1. US Accounting Services industry revenue: $158.4B in 2026, growing 1.8% that year (IBISWorld industry accounts).

2. US Payroll & Bookkeeping Services revenue: $80.9B in 2026, growing 3.8% (IBISWorld).

3. Combined US accounting labour market: approximately $239B in 2026 (sum of the two above).

4. Global Accounting Services revenue: $653.9B in 2026, 1.5% CAGR 2021–26 (IBISWorld).

5. Global accounting software market, 2026 estimates: $23.47B (Mordor Intelligence), $23.1B (Precedence Research), $22.72B (The Business Research Company).

6. US accounting labour is therefore roughly 10x the entire global accounting software market.

7. A fourth firm, Fortune Business Insights, sizes the same 2026 market at $8.05B — roughly one third of the other three. Its own figures do not reconcile internally, and we do not recommend citing it.

8. Intuit's FY2026 revenue: $21.4B, up 14% (fiscal year ended 31 July 2026).

9. Intuit Global Business Solutions segment alone: $12.9B, up 16% in FY2026 — more than half the figure most analysts give for the entire global accounting software market.

10. QuickBooks Online Accounting revenue growth: +23% in FY2026.

11. Published "AI in accounting" market estimates for 2026 cluster at $10.87B (Mordor) and $11.00B (Precedence), both at roughly 44.5% CAGR. The near-identical values suggest shared sourcing rather than independent corroboration.

12. Competing estimates for the same category range from $3.88B to $11.00B for 2026, with CAGRs from 25% to 45% — a spread of nearly 3x on size and 20 percentage points on growth.

---

Adoption

13. Finance leaders reporting AI use in the finance function: 59% in 2025, versus 58% in 2024 and 37% in 2023 (Gartner survey, n=183 CFOs and senior finance leaders, fielded May–June 2025). Adoption is flat year over year.

14. Gartner predicted in September 2024 that 90% of finance functions would deploy at least one AI-enabled solution by 2026 — while forecasting that fewer than 10% would see headcount reductions.

15. Organisations using AI in financial planning, reporting and commercial analysis: over 75% (KPMG Global AI in Finance, n=1,013 senior finance leaders across 20 countries, revenue ≥US$250M, fielded March 2026).

16. Accounting firms with no written AI policy: 87% (Financial Cents, n=486 North American accounting professionals, fielded 15 July – 7 August 2026).

17. Accounting professionals who never put client data into AI at all: 16% (Financial Cents, 2026).

18. A further 14% redact or anonymise client data before using AI tools (Financial Cents, 2026).

---

Return on investment

19. CFOs reporting a strong impact from AI investment: 7% (Gartner).

20. US CFOs reporting a clear, measurable return to date: 14% (RGP survey, n=200).

21. Finance teams with AI investments seeing measurable financial impact: 28% (CFO.com survey).

22. Organisations saying AI meets or exceeds ROI expectations: 71% — but only 23% say it exceeds them (KPMG, March 2026).

23. The three independent ROI figures above (7%, 14%, 28%) disagree by a factor of four, but all point the same direction: measured return trails self-reported adoption by a wide margin.

24. Reported performance gap for organisations deploying agentic AI versus those that aren't: approximately 32 percentage points on average, rising to ~40pp on forecast accuracy and ROI (KPMG, March 2026).

---

What is actually automated

25. AI use cases in production in finance functions, November 2025 (Gartner): knowledge management 49%, accounts payable automation 37%, error and anomaly detection 34%.

26. Accounts payable is the single highest-penetration operational finance AI use case in production.

27. Accountants using some form of automation (not AI-specific): 95% — payroll 47%, AP/AR 46%, data entry 43%.

28. Share of time accountants spend on compliance tasks: 62% (Intuit 2025 Accountant Tech Survey — vendor self-report).

29. Firms offering advisory services: 93%, up from 83% year over year (Wolters Kluwer).

---

Autonomy and trust

30. Organisations allowing AI agents to execute high-stakes decisions without human review: 5% (Grant Thornton 2026 AI Impact Survey, n=950 C-suite and senior leaders, fielded 23 February – 18 March 2026).

31. Organisations limiting agents to moderate-risk task automation: 60% (Grant Thornton, 2026).

32. Organisations giving agentic AI access to data and processes: 72% (Grant Thornton, 2026).

33. Accounting professionals who trust AI output enough to use it with limited review: 19%. Those who actively disagree they would trust AI with minimal oversight: 52% (Financial Cents, 2026).

34. Accounting professionals saying human judgment matters more, not less, in an AI world: 90% (Financial Cents, 2026).

35. Number of vendors we could identify shipping fully autonomous journal-entry posting to a general ledger, with no human approval gate: zero.

---

Governance and audit

36. Leaders who lack strong confidence they could pass an independent AI governance audit within 90 days: 78% (Grant Thornton, 2026).

37. Leaders who say they are well-prepared for AI privacy and security challenges: 40% (Grant Thornton, 2026).

38. Organisations with a tested AI incident response plan: 20% (Grant Thornton, 2026).

39. PCAOB workpaper assembly window under amended AS 1215: cut from 45 days to 14 days, effective 15 December 2026.

40. QC 1000, the PCAOB firm-level quality control standard requiring firms to ensure the technology they use is reliable and secure, is also effective 15 December 2026.

41. Amendments to AS 1105 (audit evidence) and AS 2301 (risk responses) covering technology-assisted analysis are fully effective for calendar-2026 audits.

42. Number of AI-specific PCAOB standards issued or planned: zero. Existing technology-neutral standards carry the obligation.

43. PCAOB Part I.A deficiency rate across all inspected firms: 39% in 2024.

44. Where AI-related audit deficiencies cluster: journal-entry testing — specifically, anomaly detection used without documented algorithm calibration, data lineage, or false-positive evaluation. This is a deficiency under AS 1105 regardless of whether the AI output was correct.

45. The IAASB approved development of non-authoritative guidance on technology and quality management in December 2025, publishing roundtable feedback from 240+ stakeholders on 10 February 2026. No new standard was proposed.

---

The profession

46. US accountants and auditors who left the workforce 2019–2022: over 300,000, a 17% decline from peak (BLS payroll data, analysed by the Wall Street Journal).

47. Accounting degrees conferred in AY 2023–24: 55,152, down 6.6% year over year. Bachelor's 40,817 (−3.3%); master's 14,335 (−15%) (AICPA 2025 Trends Report).

48. New CPA Exam candidates: 42,626 (2023) → 28,082 (2024) → 16,448 (first half of 2025).

49. Students enrolled in 2- and 4-year accounting programmes, Spring 2025: 266,506, up 12.4% year over year — the highest since 2020, and the second consecutive semester of roughly 12% growth (AICPA).

50. Note the direction conflict in 46–49: degrees conferred are still falling while enrolment is rising. Both are true — they measure different cohorts, and enrolment is the leading indicator.

51. AICPA and NASBA approved model legislation in May 2025 creating a bachelor's (120 hours) + two years' experience + CPA Exam pathway, as an alternative to the 150-hour rule. Roughly half of US jurisdictions now have an active alternative pathway, with most of the remainder enacted.

52. Share of small and mid-size US CPA firms using offshore outsourcing: approximately 30–35% in 2026, up from 10–15% in 2020 (industry estimate; vendor-published, no disclosed methodology).

---

Consolidation

53. Private-equity-backed accounting firm deals: 22 (2023) → 65 (2024) → 104 (2025) → 25+ in January 2026 alone.

54. Financial sponsors as a share of US accounting M&A: 49% in the twelve months to March 2026.

55. Private equity now holds stakes in 11 of the 30 largest US accounting firms, with over $50B deployed across six years.

56. Between 2015 and 2025, 177 direct PE investments in accounting firms spawned 875 bolt-on acquisitions — 1,052 transactions in total.

57. Thrive Holdings, an AI-native accounting roll-up, raised $2B in August 2026 at a $12B valuation (SoftBank, D1 Capital, Altimeter). It owns 50+ accounting firms and 2,000+ professionals. OpenAI took an equity stake in December 2025.

58. Botkeeper, an AI bookkeeping platform, shut down on 7 February 2026 after eleven years and roughly $90M raised. Its CEO attributed the closure to accounting-industry consolidation eliminating its largest clients.

---

How to use these numbers

Three cautions, because this category is unusually noisy:

Market-sizing reports are models, not measurements. Items 5, 7, 11 and 12 come from firms that sell the reports the figures appear in. Within the same category and the same year, estimates vary by up to 3x. The genuinely measured figures on this page are the IBISWorld industry accounts (1–4) and Intuit's audited results (8–10).

Self-report is not outcome. Items 13–18 are what people say they do. Items 19–23 are what they can demonstrate. The gap between the two is the most important pattern on this page.

Check the fielding date, not the publication date. Several widely cited 2026 statistics were fielded in mid-2025. We've given the fieldwork period wherever the source disclosed it.

One figure we see repeated that you should not use: a claim of "653,408 licensed accountants, down from a peak of 1.93 million in 2019." It conflates licensed CPAs with total BLS accountant and auditor employment — two different populations.

---

Have a figure we've missed, or a correction? We maintain this page — tell us and we'll update it with attribution.