KPMG: Enterprise AI Agents Hit 62% Adoption as Budgets Overshoot

KPMG's latest quarterly survey of senior business leaders paints a portrait of enterprise AI at a turning point: adoption of Enterprise AI Agents is accelerating fast, but the money to fund them is running out faster.

The KPMG Q3 2026 AI Pulse Survey, based on responses from 2,131 senior business leaders across 20 countries in September 2026, found that 62% of organizations are currently building, testing, or deploying Enterprise AI Agents — up from 53% just one quarter earlier, according to KPMG's published survey findings. Employee use of Enterprise AI Agents is climbing too: 34% of organizations reported significant employee adoption, up from 25% in Q1.

The multi-agent surge

The most striking finding in the survey is how quickly single agents are giving way to coordinated fleets. A quarter of organizations (25%) have now deployed multi-agent systems — where several Enterprise AI Agents collaborate on a single task — up from just 6% across the previous two quarters, according to the report.

Nearly four in 10 respondents said they were either building or using multi-agent systems, the survey found. That represents one of the fastest enterprise technology ramp-ups in recent memory, and it signals that Enterprise AI Agents are moving from isolated pilots to orchestrated, production-grade infrastructure.

The UK data released alongside the global results shows the same momentum. KPMG's UK analysis found 35% of UK organisations reporting significant deployment of Enterprise AI Agents, up from 30% in Q1 2026, according to The Datatech Times' coverage of the UK findings. Meanwhile, 64% of UK firms now place responsibility for AI-informed decisions at C-suite level or above, compared with 54% globally.

The cost reckoning

The same survey delivered a sobering second headline: roughly 93% of participants exceeded their AI budgets, according to Autonai News' coverage of the survey. The gap between deployment speed and budget discipline is now the central tension in enterprise AI strategy.

Agent-based workflows can be five to 30 times more computationally intensive than chatbot queries, the coverage noted. That makes architectural choices — model tiering, prompt caching, context optimisation — the primary cost lever as inference spend scales, and it explains why teams deploying Enterprise AI Agents are being asked to justify every token.

Enterprises are responding with financial discipline. KPMG reported that 74% of leaders now require cost reviews during project approval, up from 61% last quarter. Another 43% enforce token or compute usage budgets to manage spending on Enterprise AI Agents.

The pressure will only intensify. Gartner has projected that agent inference costs could rise fivefold by 2028, and the consultancy forecasts that more than 40% of agentic AI projects will be cancelled by the end of 2027, citing unclear value propositions and governance failures. McKinsey's 2026 research made the same point from the other direction: deploying Enterprise AI Agents without defined outcomes is itself a cost.

Governance separates winners from pilots

The survey reinforces a theme that has run through every major enterprise AI report this year: governance is the difference between experimentation and production. Seventy percent of enterprises now use real-time monitoring dashboards to assess AI performance, and 49% have defined high-risk scenarios where autonomous decision-making by Enterprise AI Agents is prohibited.

A separate Delinea Identity Security Report titled "The AI Enforcement Gap" put a precise number on the disconnect. According to the report, 99.7% of IT and security leaders report having a formal policy for AI agent data access and isolation, but those policies are frequently not enforced through technical controls. Policy without enforcement means Enterprise AI Agents can retain network and data access beyond their intended scope.

The cybersecurity angle is growing in priority. Coverage of KPMG's findings by Cybersecurity Dive noted that 43% of organizations include security and identity features in their AI control harnesses, followed by data access controls and monitoring of AI outputs. Nearly a quarter of organizations hold the CEO or executive committee directly responsible for AI governance — a sign that oversight of Enterprise AI Agents has become a board-level concern.

What comes next for enterprise AI agents

KPMG's data suggests the enterprise agent era has moved past the question of whether to adopt and into the harder question of how to govern, secure, and pay for what has been adopted. With multi-agent deployments quadrupling in under a year and 59% of business leaders expecting measurable ROI on their AI investments within 12 months, the pressure to turn pilot-scale experiments into governed, budgeted production systems has never been higher.

The gap between ambition and control remains the story to watch. Enterprises that pair deployment speed with management discipline — cost reviews, token budgets, named executive accountability — are the ones most likely to still be scaling when Gartner's cancellation forecast arrives. For more on how agents are reshaping the digital landscape, see AI Agents Outnumber Humans on the Web, Says Cloudflare and browse the latest coverage on genznewz's AI News topic page.

Sources

This article drew on KPMG's AI Quarterly Pulse Survey and the KPMG Global AI Pulse Q3 2026 report (PDF), with additional reporting from Autonai News, Cybersecurity Dive, and The Datatech Times.