AI Agent Adoption Statistics (2026): Deployment, Scaling and Cancellation Data for ERP Teams

Agent adoption numbers in 2026 range from 17% to 88%, and the spread depends almost entirely on what each survey counts as "using agents." This roundup separates the measures, dates each one, and shows where SAP-specific data points in a different direction from the broader market.

Key takeaways

Adoption: how many organizations use agents

The most-quoted adoption figures measure different rungs of the same ladder. Read the question before you read the number.

• McKinsey reports that 88% of survey respondents say their organizations regularly use AI in at least one business function, and 72% report using generative AI. Respondents to the McKinsey State of AI survey were collected from 1,993 participants across 105 nations between late June and late July 2025.
• Stanford's 2026 AI Index, which draws on McKinsey data, puts AI use at 88% of surveyed organizations in 2025, up from 78% in 2024.
• Gartner's 2026 CIO and Technology Executive Survey is the strictest measure. It finds that only 17% of organizations have deployed AI agents to date, yet more than 60% expect to do so within the next two years, per its 2026 Hype Cycle for Agentic AI.

Source: McKinsey, The State of AI, Nov 2025 (survey fielded mid-2025)

The same survey shows the breadth problem. In any given business function, no more than 10 percent of respondents say their organizations are scaling AI agents. Read that bar as a cap, not an average.

Source: Gartner, 2026; McKinsey, Nov 2025; Deloitte, 2026; KPMG, Q3 2026; LangChain State of Agent Engineering, Dec 2025, via Prefactor. Definitions differ; do not read this as a ranking.

Two things to note. The LangChain figure covers respondents who build agents, so it is a practitioner sample and not a general-enterprise one. The KPMG figure uses a definition that changed between quarterly releases (see the conflicts section below).

Scaling and the pilot gap

• McKinsey's respondents say nearly two-thirds of organizations have not yet begun scaling AI across the enterprise.
• Only about 39% report enterprise-level EBIT impact from AI, and that impact is typically under 5%, according to McKinsey's summary of the 2025 survey.
• McKinsey's high performers, roughly 6% of respondents, treat AI as transformation, redesign workflows, show visible leadership ownership, use human-in-the-loop governance, and scale agents across more functions.
• Among agent builders in LangChain's December 2025 survey, 57% have agents running in production, up from 51% a year earlier, rising to 67% at organizations with more than 10,000 employees.
• KPMG's Q2 2026 release says 53% of organizations are using AI agents this quarter, compared to 55% last quarter.

Source: KPMG AI Quarterly Pulse, Q3 2026 (large US organizations, $1B+ revenue; KPMG reports the prior two quarters together)

Forecasts: where adoption is heading

Forecasts are not outcomes. Check each one against its date, and check it again when the target year arrives.

• Gartner expects that 40% of enterprise applications will be integrated with task-specific AI agents by the end of 2026, up from less than 5% today. That deadline is about seven weeks away.
• By 2028, Gartner expects that 33% of enterprise software applications will include agentic AI, up from less than 1% in 2024.
• Gartner also predicts that at least 15% of day-to-day work decisions will be made autonomously through agentic AI by 2028, up from 0% in 2024.
• Deloitte's 2026 report, based on a survey of 3,235 leaders fielded in August and September 2025, expects that 74% of respondents expect their companies to be using AI agents at least "moderately" by 2027.

Source: Deloitte, State of AI in the Enterprise 2026 (3,235 leaders, Aug–Sep 2025)

Value and ROI

PwC's May 2025 survey is the most cited source for agent-specific value. Its sample is US executives, and its numbers describe what adopters report, not what all companies achieve.

• Of the executives PwC surveyed, 79% say their organizations are already using AI agents. PwC's AI Agent Survey was fielded in spring 2025.
• Among companies already using agents, 66% report increased productivity, 57% report cost savings, and 54% report improved customer experience.
• 88% say their team or business function plans to increase AI-related budgets in the next 12 months due to agentic AI.
• Only 45% are fundamentally rethinking operating models around agents, and 42% are redesigning processes around AI agents.

Source: PwC, AI Agent Survey, May 2025

Source: PwC AI Agent Survey, May 2025, as summarized by theCUBE Research; base is adopters only.

Measurement is a weak point across the market. KPMG's Q2 2026 release reports that only 26% of organizations have full, real-time visibility into what their AI systems cost to operate, even though most have monitoring dashboards and approval processes in place.

Failure and risk

• Gartner predicts that over 40% of agentic AI projects will be canceled by the end of 2027, due to escalating costs, unclear business value or inadequate risk controls.
• Gartner also states that most deployments remain narrowly scoped, and fully autonomous agents are not ready for the majority of enterprise use cases.
• Deloitte's survey finds that only 21% of respondents say their organizations have a mature governance model in place for agentic AI.
• KPMG's Q3 2026 release reports that 80% of banks cited data readiness and access as the top challenge to AI agent deployment, up from 63% last quarter. That figure covers banking respondents only.

Source: Deloitte, State of AI in the Enterprise 2026

SAP and Joule: the ERP-specific picture

SAP-specific adoption data is thinner than the market data, and the two main datasets disagree.

• The DSAG Investment Survey 2026 found that only 3% of SAP customers run SAP Business AI in production, while 77% of AI-active enterprises rely on non-SAP tools like Microsoft Copilot.
• SAP has disputed the survey's representativeness. According to coverage in Techzine, SAP's Chief Customer Officer argued that fewer than 100 organizations participated in the survey.
• In the same coverage, SAP cited that 68 percent of Fortune 500 customers use SAP AI, but this does not specifically refer to Joule.
• Futurum's enterprise decision-maker survey (n=833) reports that 73.1% of enterprise technology decision makers rank agentic AI among their top technology priorities.

Source: Techzine, DSAG Investment Survey 2026 and SAP response, 2026. These bases differ, and the two bars should not be compared directly.

Source: Techzine on DSAG Investment Survey, 2026

SAP's own release notes report customer productivity results. These are vendor-published and have not been independently verified. SAP's Q2 2026 release notes say that developers saw a 20% increase in productivity using Joule to automate routine coding tasks and optimize code, citing Bosch Digital. The same notes say Joule also generates test cases, speeding up unit testing by 15% to 20%. For HR, LC Waikiki used an AI agent built on SAP Joule to cut HR process cycle times by 40% to 60%, per SAP's report. Source: SAP News Center, AI Release Highlights Q2 2026.

Testing and quality

We found few sourced statistics that measure AI agents in ERP testing specifically. The closest data points are the SAP-published unit-testing figure above and the broader governance figures. Teams should be careful about extending general enterprise data to test-automation decisions. Test outcomes have not been measured in the surveys we reviewed, so the lack of a number is itself a finding.

• Deloitte's respondents are split on oversight: companies succeeding with agentic AI are starting with lower-risk use cases, building governance capabilities, and scaling deliberately.
• KPMG reports that two-thirds of organizations have monitoring dashboards (66%) and approval processes (61%) in place.

People, skills and budgets

• Deloitte reports that workforce access to AI has expanded by 50% in just one year, growing from fewer than 40% to around 60% of workers equipped with sanctioned AI tools.
• KPMG's Q2 2026 release says employee adoption of AI agents increased quarter over quarter to 56% from 23% in Q1 2026.
• In PwC's survey, 48% of surveyed senior executives said that they will likely increase headcount due to the change AI agents will bring to how we work.

Where the numbers conflict

• Definitions drive the spread. Gartner counts deployed agents. McKinsey counts scaling in at least one function. KPMG counts a deployment rate that has changed across releases. Deloitte counts use at least moderately. Each number is accurate for its question.
• KPMG's own series is inconsistent. A KPMG page reports deployment increased from 53% to 62% since last quarter, while another KPMG page says agent deployment more than doubled from 11% in Q1 to 26% in Q4. Q3 2026 then reports AI-agent deployment increased to 30% in Q3, up from 19% in Q2. We did not reconcile these and did not build a trend line from them.
• SAP's Joule numbers conflict. The DSAG figure (3%) and SAP's Fortune 500 figure (68%) use different populations and different products, as shown above.
• PwC sample size. PwC's own page describes the 300 senior executives in our May 2025 survey, while coverage of the same survey cites 308 U.S. respondents. Use the PwC page as the authoritative source.
• Large-company subgroup. Summaries of McKinsey's 2025 data describe larger companies (revenue over $5B) as more likely to be scaling, with nearly half in the scaling phase. A separate secondary summary cites 40% of respondents at organizations with over $1 billion in revenue scaling agents in at least one function, versus 22% at smaller ones. We could not verify either subgroup figure against McKinsey's report, so we did not use them.
• Excluded. Several aggregator sites cite figures such as 51% in production, 85% planning agents, and an 80% Gartner Q1 2026 survey result. We could not trace these to a primary publisher, so they are not in this article.

What this means for your sign-off

• Ask which definition a vendor or analyst uses. A "deployed" figure of 17% and a "scaling" figure of 23% can both be true. Your business case should state which one it relies on.
• Treat SAP's Joule data as unsettled. Before you approve an agent for a sandbox or a test run, confirm which Joule capabilities are generally available for your release and tier. Our SAP AI for practitioners guide covers what business AI, Joule and agents mean for an S/4HANA program.
• Put governance ahead of agent coverage. When only about one in five organizations has a mature governance model, the sign-off question is whether a human approves agent-generated test cases, scripts and data changes. The AI Agent Guardrails and Approval Gate Register and the AI-Generated Test Review Checklist are built for that step.
• Test AI-generated tests like any other change. Our review of Claude and Playwright for SAP testing compares AI-assisted scripting with established SAP test tools, including where it breaks down.
• Measure cost before you scale. Given that few organizations can see AI operating costs in real time, add per-run cost and usage limits to your test-automation business case. Our regression automation priorities guide is a good starting point for deciding where automation pays first.

Frequently asked questions

What percentage of companies use AI agents in 2026?

It depends on the definition. Only 17% of organizations have deployed AI agents to date, according to Gartner's 2026 CIO survey. McKinsey's 2025 survey found that 23% of organizations are actively scaling an agentic AI system in at least one business function.

How many agentic AI projects will be canceled?

Gartner predicts that over 40% of agentic AI projects will be canceled by the end of 2027. This is a forecast, not a measured outcome.

How many SAP customers run Joule in production?

The DSAG Investment Survey 2026 reports 3% of SAP customers running SAP Business AI in production. SAP disputes the survey's representativeness, and the data is not Joule-only.

How many organizations have mature agent governance?

Deloitte's 2026 report found that only 21% of respondents say their organizations have a mature governance model in place for agentic AI.

Do agents reduce testing time in SAP projects?

The only sourced figure is from SAP. SAP's Q2 2026 release notes say Joule-generated test cases speed unit testing by 15% to 20%, reported by a customer and not independently verified. We found no independent measurement of agent-driven test cycle time.

How many organizations scale agents across functions?

McKinsey's 2025 data shows that in any given business function, no more than 10 percent of respondents say their organizations are scaling AI agents. KPMG's Q3 2026 figure for multi-agent systems is 25%, up from 6% in the prior two quarters.

Sources

1. Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027, Gartner, June 25, 2025
2. 2026 Hype Cycle for Agentic AI, Gartner, 2026
3. The Latest Gartner Hype Cycle for Artificial Intelligence, Gartner, 2026
4. Gartner Predicts 40% of Enterprise Apps Will Feature Task-Specific AI Agents by 2026, Gartner, August 2025
5. The State of AI: Global Survey 2025, McKinsey, November 2025
6. PwC's AI Agent Survey, PwC, May 2025 survey
7. Scaling Enterprise Operations with PwC's AI Strategy, agent OS, and Secure Agentic Architectures, theCUBE Research, summarizing PwC data, 2025
8. AI Investment and Agent Deployment Hold Steady Amid Growing Focus on Pragmatism, KPMG, June 24, 2026
9. AI's Value Story Sharpens as Organizations Gain Confidence in Governance, Accountability and Workforce Adoption, KPMG, Q3 2026
10. KPMG AI Quarterly Pulse Survey, KPMG, Q4 2025 results published January 2026
11. Business and IT leaders report AI agents are scaling faster than their guardrails, Deloitte, 2026
12. The State of AI in the Enterprise, Deloitte AI Institute, 2026
13. SAP blocks external AI agents. Salesforce and ServiceNow don't., Techzine Global, 2026 (reporting DSAG Investment Survey 2026)
14. SAP Business AI: Release Highlights Q2 2026, SAP News Center, July 2026
15. SAP's Autonomous Enterprise: Is Joule the ERP Endgame?, Futurum Group, 2026
16. 79% of Companies Run AI Agents: 65 Adoption Stats (2026), Prefactor, summarizing LangChain State of Agent Engineering, December 2025 (secondary)
17. Enterprise AI Statistics 2026: Adoption, ROI & Deployment, summarizing Stanford HAI AI Index 2026 (secondary)

Last updated: November 8, 2026. Statistics are refreshed as new research is published.