Most companies are spending enormous sums on artificial intelligence and getting surprisingly little back — and a major new global study from KPMG explains exactly why. The findings from KPMG’s first quarterly Global AI Pulse survey are both clarifying and uncomfortable: the gap between what enterprises are investing in AI and what they are actually gaining from it is not closing. For most organizations, it is widening.
The survey captures a moment that feels familiar to anyone watching enterprise technology cycles. Boards have approved the budgets. Vendors have been selected. Pilots have run. And yet, only 11 percent of organizations have reached the stage where AI agents are deployed at scale and producing measurable, enterprise-wide business outcomes. That number is the one that should be driving conversations in every C-suite right now.
The $186 Million Question Nobody Is Asking Correctly
The average global organization plans to spend $186 million on AI over the next 12 months. That figure sounds like serious commitment. But investment size, it turns out, is almost entirely decorative as a predictor of AI performance. KPMG’s data makes this plain: spending more does not equal creating more value. The organizations extracting the most from AI are not necessarily the biggest spenders — they are the ones who changed their approach before they changed their tools.
Regional breakdowns add texture to this picture. Asia-Pacific organizations lead spending at $245 million on average, followed by the Americas at $178 million and EMEA at $157 million. Yet regional spend rankings do not map neatly onto outcome rankings. The variable that predicts performance is not budget size — it is deployment philosophy.
What “Meaningful Business Outcomes” Actually Means
Here is where the KPMG data requires careful reading. Sixty-four percent of survey respondents say AI is already delivering meaningful business outcomes. On the surface, that sounds like success. But “meaningful” is doing enormous work in that sentence, and the distance between a productivity nudge and a genuine margin gain is vast.
Think of it this way: if a company installs an AI tool that helps a finance analyst summarize reports 40 percent faster, that is meaningful to that analyst. It is not, however, restructuring how financial decisions flow across the enterprise. The first is an efficiency tweak. The second is what KPMG’s report calls the territory of “AI leaders” — the 11 percent operating at a fundamentally different level.
The Architecture of the Performance Gap
KPMG’s report draws a sharp distinction between organizations that are scaling agentic AI and everyone else. Among those AI leaders, 82 percent report meaningful business value from their AI investments. Among the remaining majority, that figure drops to 62 percent. A 20-percentage-point gap sounds manageable until you understand what it actually reflects.
The leading organizations are not just using better AI models. They have deployed agents that coordinate work across departments, route decisions without requiring human approval at every single step, surface operational insights in near real-time, and catch anomalies before they become crises. In IT and engineering, 75 percent of AI leaders use agents to accelerate code development versus 64 percent of their peers. In supply chain operations, it is 64 percent versus 55 percent. These gaps compound over time.
Process Redesign First — Then AI Deployment
The single most important insight from this research is about sequencing. Most enterprises have layered AI onto existing workflows — a summarization tool here, a co-pilot there — without redesigning the underlying process those tools sit inside. The result is incremental improvement built on a foundation that was never optimized for AI-era operations.
The organizations pulling ahead have inverted this entirely. They redesign the process first, then deploy agents to operate within the redesigned structure. It is the difference between adding a faster engine to a cart and building a car from scratch. Both involve engines. Only one produces a structurally different outcome.
Over a three-to-five-year horizon, this distinction in approach is likely to be the defining competitive variable across industries including financial services, logistics, healthcare administration, and enterprise software.
Governance and Trust: The Hidden Bottleneck
Scaling AI agents across an enterprise is not purely a technical challenge. KPMG’s data consistently surfaces governance and trust as the friction points holding most organizations back. Agentic AI — systems that take autonomous actions, make sequential decisions, and operate across organizational boundaries — requires a fundamentally different governance framework than a chatbot or a dashboard.
When an AI agent can route a procurement decision, flag a compliance issue, or trigger a customer service response without a human in the loop at each step, the question of accountability becomes urgent. Organizations that have solved this problem — defining clear guardrails, establishing audit trails, and building internal trust in agent behavior — are precisely the ones operating at scale. Those still debating governance are still running pilots.
Key Findings from KPMG’s Global AI Pulse Survey
| Metric | AI Leaders (Top 11%) | All Other Enterprises |
|---|---|---|
| Reporting meaningful AI business value | 82% | 62% |
| Using AI agents for code development | 75% | 64% |
| Using AI agents for supply chain operations | 64% | 55% |
| Average planned AI spend (global) | $186 million over next 12 months | |
| ASPAC average planned spend | $245 million | |
| Americas average planned spend | $178 million | |
| EMEA average planned spend | $157 million | |
| Enterprises scaling agentic AI enterprise-wide | Only 11% | |
What the Next 12 to 24 Months Actually Look Like
The KPMG data signals something important about the near-term trajectory of enterprise AI. The organizations currently in that 11 percent are not just ahead — they are building structural advantages that will be very difficult for laggards to close once those advantages start compounding at scale. Margin improvements, faster decision cycles, and reduced operational overhead create resources that can fund further AI investment, creating a self-reinforcing loop.
For the remaining 89 percent, the strategic window is not closed but it is narrowing. The organizations most likely to move up are those willing to question their deployment sequencing — asking not “where can we add AI?” but “which processes should we redesign entirely, and how should AI operate within the new structure?” That shift in framing, more than any new model release or vendor announcement, is what separates the leaders from the rest.
If you are tracking enterprise AI strategy — whether as a business leader, an investor, or simply someone trying to understand where this technology is actually going — the KPMG Pulse survey is one of the clearest maps available right now. I will be watching the Q2 results closely to see whether that 11 percent figure starts to move. When it does, that will be the real signal that enterprise AI has crossed a meaningful threshold. Stay with us at STI2 as we continue to track that shift.