AI in ERP Statistics (2026): SAP Adoption, Agentic AI, Testing and Risk Data

SAP customers and the wider ERP market are moving from AI pilots toward production, but the sources disagree on how far along they are. This roundup collects sourced figures from SAP, the SAP user groups, SAPinsider, Gartner, Capgemini and testing-industry surveys published through October 2026. Each figure names its year, its population and its source.

Key takeaways

Where SAP customers stand on AI

Most SAP customers are still early in their AI programs. SAPinsider's 2026 benchmark found that 74% of SAP customers remain in experimentation, and only 16% of technology leaders use AI beyond a limited scope. In the same research, only 18% of organizations had reached an integrated stage with AI embedded in several processes, and 14% described themselves as optimized or transformational.

The DSAG 2026 investment report points to a similar picture from a different angle. DSAG respondents report that 43% of SAP customers have already implemented AI use cases, and the same report notes that many of those customers are buying AI tools from outside SAP. Innobu, summarizing the DSAG findings, puts the share running SAP Business AI in production at 3%, and says 77% of AI-active enterprises rely on non-SAP tools such as Microsoft Copilot or ChatGPT.

Source: SAPinsider, 2026; DSAG Investment Report 2026, via Constellation Research, Feb 2026; DSAG 2026, via Innobu. The three measures use different definitions and should not be read as one funnel.

ASUG's February 2026 research on AI adoption asked 142 members, and the study was sponsored by SAP, Microsoft and Intel. It found 41% actively piloting AI, 39% building foundational knowledge, and 24% in active deployment.

Source: ASUG 2026 SAP AI research, via CIO, May 2026; sample of 142 ASUG members, per TechTarget, Aug 2026.

Older user-group data shows the same slow start. An earlier DSAG, ASUG and UKISUG survey, which our source does not date, found that only 6%, 7% and 5% of respondents respectively were using AI in many areas (historical).

• Historical: DSAG 6%, ASUG 7%, UKISUG 5% using AI in many areas (undated in source).
• DSAG's chairman said that almost a third of companies still had no official AI plans in that earlier survey (historical, undated in source), per the same CIO article.

The SAP Business AI platform: what SAP reports

SAP's own figures describe a fast-growing product catalog. These are vendor data and measure SAP's output, not customer usage.

• SAP reported more than 30 specialized agents and over 2,500 Joule Skills in its Q1 2026 release highlights, with Joule live across 35 solutions (SAP, vendor data).
• SAP's Q3 2026 highlights include an estimated 35% reduction in the time to build and deploy custom Joule Agents. This is an SAP estimate, not a measured customer result.
• SAP's Sapphire 2026 announcement set aside a €100 million fund for partners to help customers deploy SAP-built assistants and agents, and RISE with SAP customers get three assistants activated in their first year (SAP, vendor data).
• Innobu reports that Joule is available only to customers on RISE with SAP or GROW with SAP contracts. Our search did not find an SAP primary source for that eligibility rule, so check your contract terms before you plan around it.

An SAP-commissioned survey, reported by Quantumrun, puts the average return on AI investments at 16% among 1,600 executives across eight countries, with respondents expecting that to roughly double within two years. We could not check the original SAP and Oxford Economics report, so treat the figure as secondary.

Migration and budgets: AI is now part of the business case

AI has moved from a side topic into the reason many customers give for migrating. SAPinsider's 2026 benchmark found that 43% of organizations named SAP's AI announcements as the main external factor shaping their ERP strategy, ahead of the 2027 maintenance deadline at 39%.

Source: SAPinsider ERP Migration and Transformation 2026 Benchmark, 2026. Respondents were SAP community members.

The migration itself is moving, though not quickly. ASUG data shows that 56% of customers are either live on S/4HANA or have started their migration, up from 45% in 2024. The ASUG 2026 Pulse survey also found that budget constraints are the top challenge, cited by 61% of respondents, 7 percentage points higher than the year before.

Source: ASUG data, via CIO, May 2026.

DSAG's 2026 investment report adds that customers are being more selective with spending, moving slowly to S/4HANA, and often extending maintenance on on-premises systems. The same report notes that its 2026 survey reflects more large enterprises than earlier editions, which affects comparisons with older DSAG data.

Spend and market forecasts

Analyst forecasts are the clearest signal of where vendors expect money to flow, but they are projections rather than measured results.

• Gartner's February 2026 forecast puts worldwide spending on ERP software with agentic capabilities at $72 billion by 2030, a five-year compound annual growth rate of 74.6%.
• Gartner also says that widescale adoption of agentic capabilities will be constrained by ecosystem readiness, and that the constraint varies by ERP subsegment over the next five years (same Gartner document).
• Gartner's Predicts 2026 research on ERP, as excerpted by Rimini Street, expects only 30% of organizations to have sufficient data quality to fully use advanced AI by 2027.
• The same Gartner excerpt predicts that over 70% of acquired ERP AI capabilities will become shelfware by 2030.
• A Rambase summary attributes to Gartner a forecast that more than half of routine ERP tasks will be executed autonomously by AI by 2030. We could not find the primary Gartner text for this line, so treat it as secondary.
• Gartner's June 2025 release says at least 15% of day-to-day work decisions will be made autonomously by 2028, up from 0% in 2024, and that 33% of enterprise software applications will include agentic AI by 2028.

Risk, governance and failure

Gartner's June 2025 prediction is the most-cited failure forecast in this list. It names three causes: escalating costs, unclear business value and inadequate risk controls. Gartner also estimates that only about 130 of the thousands of agentic AI vendors offer real agentic capabilities. Secondary write-ups describe the underlying poll differently: one says it covered more than 3,400 organizations, and another describes a January 2025 poll of 3,412 webinar attendees. Check the methodology in Gartner's release before you quote a sample size.

Security teams are less comfortable than the business. Onapsis's June 2026 study of 204 senior cybersecurity leaders in U.S. organizations found the following.

Source: Onapsis State of AI, Security and ERP report, via SAPinsider, June 2026.

• SmartBear's September 2026 survey of 1,436 U.S. leaders and practitioners found that 46% of teams have shipped AI-generated code that later failed in production.
• The same survey found 69% still confident in AI-written code; the headline figure comes from the press release title.
• SAPinsider's 2026 data identifies accuracy and reliability as the top AI concern for 63% of SAP customers.
• Google Cloud's 2025 DORA report found that 30% of respondents report little or no trust in AI-generated code, a slightly lower share than the year before.
• The same DORA report links AI adoption to higher software delivery throughput but also to lower delivery stability.

AI in testing and quality engineering

Testing adoption figures are high, but the definitions behind them vary widely. The World Quality Report 2025 gives a more conservative view of enterprise scale.

BrowserStack's 2026 State of AI in Software Testing, which surveyed 250-plus CTOs, VPs of Engineering and QA leaders in the US, UK and Europe, found that 94% of teams use AI in testing and 61% use it across most workflows. The same survey found that only 12% of teams have reached full AI autonomy.

Source: BrowserStack, PractiTest and Katalon figures, as compiled by TestDino, 2025-2026; Rainforest QA State of Test Automation, Jan 2026, as compiled by Quash; Leapwork survey, 2026, via DevOps.com. Samples and definitions differ; the bars are not a ranking of equal measures.

Leapwork's survey of more than 300 engineers and IT decision-makers found that 65% are experimenting with or using AI in at least some testing, but only 12.6% have embedded it across core test workflows.

The Capgemini-led World Quality Report 2025, the 17th edition, is the most detailed view of enterprise scale. It reports that nearly 90% of organizations are pursuing generative AI in quality engineering, but only 15% have reached enterprise-scale deployment.

Source: World Quality Report 2025, Capgemini, Sogeti and OpenText, Nov 2025.

• Organizations report an average productivity gain of 19% from Gen AI in quality engineering, but one third have seen minimal gains.
• A sector excerpt from the same report says only 6% of enterprises use Gen AI in ERP testing. This is a sector-level figure from the Technology, Media and Telecommunications edition, so confirm the full-report base before quoting it.
• Synthetic data use in testing rose from 14% in 2024 to an average of 25% in 2025, and the report ranks synthetic data as the top Gen AI use case.
• The same report says 60% of organizations struggle to provide secure, scalable test data.

Source: World Quality Report 2025-26, Capgemini.

On the developer side, Google Cloud reports that 90% of survey respondents use AI at work, across nearly 5,000 technology professionals. That is the broadest adoption figure in this roundup, but it measures general AI use, not testing.

Source: Google Cloud, 2025 DORA State of AI-assisted Software Development, Sept 2025.

People and skills

• Half of organizations surveyed in the 2025 World Quality Report say they lack AI/ML expertise, unchanged from 2024.
• Generative AI ranks as the top skill for quality engineers, cited by 63% of respondents, ahead of core quality engineering skills at 60%.
• The 2025 report lists integration complexity (64%), data privacy risks (67%) and hallucination and reliability concerns (60%) among top challenges.

Source: World Quality Report 2025, Capgemini, Sogeti and OpenText, Nov 2025. Respondents could select multiple challenges.

Where the sources disagree

Several headline numbers in this roundup conflict. The gaps usually come from different questions, not from errors, but you should know which question a number answers before you use it in a sign-off deck.

• AI in testing: BrowserStack reports 94% of teams using AI in testing, while Leapwork reports 65% experimenting or using AI in some testing and 12.6% embedded across core workflows. The first counts any use. The second counts experimenting with or using AI in some activities and then separately counts embedded use.
• SAP customers and AI: DSAG's 43% who have implemented AI use cases sits alongside SAPinsider's 74% still experimenting and ASUG's 24% in active deployment. The populations and definitions differ, and DSAG's 2026 survey includes more large enterprises than earlier editions.
• Gartner agentic cancellations: the 40% forecast is a prediction, not an observed count. Secondary summaries disagree on the poll's base, as noted above.
• DORA adoption change: Google reports 90% adoption as a 14% increase from last year. Our sources do not say whether that increase is in percentage points or a relative change, so we do not compute one.

What this means for your sign-off

• Define which AI is in scope. Separate SAP Business AI features from third-party copilots and tools in your test scope. The DSAG data suggests many AI-using customers rely on tools outside SAP, which means interfaces and data flows you may not own. Our SAP AI for Practitioners guide covers what Joule, agents and the AI platform mean for the program.
• Treat data quality as an entry criterion. If Gartner's 30% data-quality readiness figure is even directionally right, AI features will underperform on migrated master data. Plan test-data copies, masking and synthetic data as part of the migration test plan, starting with our S/4HANA test data guide.
• Keep a human in the review loop. SmartBear reports that 84% of respondents use at least one form of human review on AI-generated tests, and only 3% rely on AI self-validation alone. Use an AI-generated test review checklist and an agent approval gate register in your evidence pack. Both are in the template library.
• Set approval gates before agents touch sensitive data. With 57% of security leaders reporting at least one business unit objecting to agent access, expect access and approval questions at sign-off. Document which agent actions need human approval and who holds that decision.
• Justify the spend with measured results. With budget constraints the top challenge for 61% of ASUG respondents, and the World Quality Report finding that a third of organizations saw minimal productivity gains, your business case needs baseline metrics. Our Joule prompts guide covers how consultants use Joule, which can help you plan a pilot with a defined measure.
• Plan for vendor cadence, not one-time go-live. Agent capabilities change with SAP's release cycle. Test each release window for AI behavior changes, not only for transactional regression.

Frequently asked questions

How many SAP customers run SAP Business AI in production?

According to the DSAG Investment Survey 2026, as reported by Innobu, 3% of surveyed SAP customers run SAP Business AI in production. Other DSAG findings put the share with implemented AI use cases at 43%, so the production figure is much narrower.

What share of QE organizations use generative AI?

The World Quality Report 2025 found that 89% of responding organizations are piloting or deploying Gen AI-augmented workflows, with 37% in production.

Will agentic AI projects be canceled?

Gartner predicted in June 2025 that over 40% of agentic AI projects will be canceled by the end of 2027. It is a forecast, not a measured outcome.

How often does AI-generated code fail in production?

SmartBear's 2026 survey of 1,436 U.S. leaders and practitioners found that 46% of teams have shipped AI code that later failed in production.

How many SAP customers are migrating or live on S/4HANA?

ASUG data reported by CIO shows 56% of customers live on S/4HANA or having begun migration, up from 45% in 2024.

Are AI announcements now a bigger factor than the 2027 deadline?

In SAPinsider's 2026 benchmark, 43% of organizations cited SAP's AI announcements as the primary external factor, ahead of the maintenance deadline at 39%.

Sources

1. SAP LeanIX Agentic AI Survey 2026, SAP LeanIX, March 2026 fieldwork
2. SAP Industry AI: Why Context Defines the Autonomous Enterprise, SAPinsider, 2026
3. The State of AI and ERP Security: 2026 Threat Research, SAPinsider on Onapsis research, June 2026
4. SAPinsider Research: 5 Hard Truths About the 2026 SAP ERP Migration Landscape, SAPinsider, 2026
5. SAP customers say migration is eating their budgets, and AI is next in line, CIO, May 2026
6. SAP AI customers link scale to governance, security, operational uses, TechTarget, Aug 2026
7. DSAG: SAP customer spending selective, AI drags, Constellation Research, Feb 26, 2026
8. SAP Joule 2026: Agentic AI, Promise vs. Reality, Innobu, 2026 (secondary, summarizing DSAG 2026)
9. SAP customers still slow to deploy AI broadly, CIO, undated in source (historical)
10. Forecast Analysis: Agentic AI in ERP Software, 2026, Gartner, Feb 5, 2026
11. Predicts 2026: The Future of ERP, Gartner excerpt republished by Rimini Street, 2026
12. Gartner Predicts 2026: The Future of ERP, Rambase, 2026 (secondary)
13. Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027, Gartner press release, June 25, 2025
14. Gartner: 40% of agentic AI projects will fail, making humans indispensable, MarTech, 2025 (secondary, for poll base)
15. Why Agentic AI Projects Get Canceled (and How to Ship), Digital Applied, 2026 (secondary, for poll base)
16. SAP Statistics 2026: Enterprise Software Adoption And Cloud Transition Data, Quantumrun, 2026 (secondary)
17. SAP Business AI: Release Highlights Q1 2026, SAP News, April 2026
18. AI Release Highlights Q3 2026, SAP News, October 2026
19. SAP Unveils the Autonomous Enterprise, SAP News, May 2026
20. World Quality Report 2025-26, Capgemini, 2025-26 edition
21. World Quality Report 2025: AI adoption surges in Quality Engineering, but enterprise-level scaling remains elusive, Capgemini, Nov 13, 2025
22. World Quality Report 2025: Quality Engineering AI Adoption, Software Testing Magazine, Nov 14, 2025
23. World Quality Report 17th Edition 2025-26, Technology, Media and Telecommunications excerpt, Capgemini and Sogeti, 2025
24. SmartBear State of Software Quality and Testing report, SmartBear, Sept 30, 2026
25. Announcing the 2025 DORA Report: State of AI-Assisted Software Development, Google Cloud, Sept 2025
26. State of AI in Test Automation 2026: Data + Reality, TestDino, 2026 (secondary, summarizing BrowserStack, PractiTest and Katalon)
27. AI Testing: Statistics, Trends & Adoption Data (2026), Quash, 2026 (secondary, summarizing Rainforest QA)
28. Survey: Adoption of AI Software Testing Slowed by Trust Issues, DevOps.com on Leapwork research, 2026

Last updated: October 31, 2026. Statistics are refreshed as new research is published.