A new KPMG survey has put a number on the gap between AI ambition and AI delivery: 11 percent.

That is the share of global enterprises that KPMG classifies as “AI leaders” in its first Global AI Pulse Survey, released March 31, 2026. These are organizations that have moved past experimentation and are actively scaling AI agents across their operations. Everyone else — the other 89 percent — is still running pilots, managing isolated deployments, or trying to figure out why their AI investments are not producing the returns they expected.

The survey draws on responses from 2,110 C-suite and senior business leaders across 20 countries. The organizations represented are not small businesses dabbling in technology. Three-quarters of respondents lead organizations with annual revenues above US$1 billion.

The investment is not the problem

The companies stuck at the 89 percent are not failing because they are being cautious with money. The survey found that global organizations plan to spend a weighted average of US$186 million on AI over the next 12 months. Seventy-four percent say AI will remain a top investment priority even if economic conditions deteriorate.

This is the uncomfortable finding at the heart of the report: most enterprises are spending serious money on AI without getting serious results. The problem is not budget. It is approach.

What the 11 percent are doing differently

The organizations in KPMG’s AI leader group share one behavior the others do not. They redesign the process first, then deploy agents into the redesigned structure. The 89 percent tend to do the opposite — they lay AI on top of existing workflows and wonder why adoption is slow and value is hard to measure.

Steve Chase, Global Head of AI and Digital Innovation at KPMG International, put it directly in the report: “The first Global AI Pulse results reinforce that spending more on AI is not the same as creating value. Leading organizations are moving beyond enablement, deploying AI agents to reimagine processes and reshape how decisions and work flow across the enterprise.”

The results speak to that difference. Among the 11 percent of AI leaders, 82 percent report meaningful business value from their AI deployments. For everyone else, that number is 62 percent. Both groups are using AI. One group is getting a meaningfully better return.

The AI leaders are also operating at a different level of integration. Their agents coordinate work across functions, route decisions without human sign-off at every step, surface enterprise-wide insights from operational data in near real-time, and flag problems before they escalate. That is not a chatbot. That is an AI workforce.

The compounding advantage problem

The survey flags something worth paying attention to. The gap between the 11 percent and the rest is not standing still. AI leaders are compounding their advantage every quarter while the majority of organizations are still sorting out foundations.

KPMG’s framing is direct: “the question for the remaining 89 percent is not whether to accelerate AI deployment, but how to do so without compounding the integration debt and governance deficits that are already constraining their returns.”

That is the real risk. The companies getting this right now are pulling further ahead. The companies running endless pilots are building technical debt and losing ground.

KPMG will release the full report at kpmg.com/aipulse on April 15, 2026. The preliminary findings are available now in the press release.

What This Means for Business

If your organization is in the 89 percent, the KPMG finding points to a specific diagnosis: the issue is almost certainly not the AI tools you are using. It is the sequence in which you are deploying them.

Agents placed on top of broken or undocumented processes do not fix those processes. They inherit them. The organizations closing the gap have done the harder work first — mapping how decisions actually flow, documenting what systems actually do, redesigning the workflow before automating it.

This is not a technology problem. It is an operations problem that technology can solve once the operations side is sorted.

For a deeper walkthrough of tools like this and how they fit together, the free Working With Claude field guide covers the ecosystem end to end. Get the guide.

The 11 percent are not smarter. They just started with the right question.