Enterprise AI is hitting a critical inflection point: organizations can build and deploy AI agents at scale, but most lack the governance and coordination to run them as integrated systems. According to KPMG's Global AI Pulse report, which surveyed 2,110 C-suite and senior business leaders at organizations with at least $100 million in annual revenue, 95% of organizations now have an AI strategy and plan to invest an average of $186 million over the next 12 months. Yet only 8% have established measurable return on investment, even though 64% reported meaningful business value .

The disconnect is striking. While 39% of organizations are scaling AI or driving adoption across the enterprise, 54% remain in research, experimentation, or strategic planning. Within that landscape, only 11% qualify as "AI leaders," defined as organizations in the top two maturity stages with the most advanced agent deployment .

Why Are Agents Deploying Faster Than They're Coordinating?

The data reveals a troubling pattern. KPMG found that 22% of organizations are still exploring AI agents, 17% are piloting them, 14% are deploying them, and 18% are scaling them across multiple functions. Another 17% are developing or implementing multi-agent systems, but only 9% have reached orchestration across workflows. Even when combining multi-agent work with broader coordination capabilities, the share rises only to 26%, leaving most organizations short of enterprise-wide orchestration .

Agents are already embedded in core business functions. KPMG reported that agentic AI is embedded in technology or IT at 66% of organizations, in operations at 55%, and in marketing and sales at 43%. More than half of respondents using agents said they are automating workflows that span multiple functions, while 41% said agents provide shared knowledge environments and 40% said they support joint decision-making across teams .

This fragmentation reflects a broader challenge facing enterprise technology leaders. According to CIO.com research, 71% of IT leaders plan to increase investment in AI-enabled technologies, while half already have AI in production in at least one business unit. However, organizations are discovering that scaling AI securely, effectively, and at enterprise level is far more complex than launching pilots. A clear divide exists between organizations realizing returns on AI and those that are not, often due to gaps in governance, data readiness, and operating models .

What Are the Real Barriers Holding Back Enterprise AI?

KPMG identified multiple obstacles preventing organizations from achieving orchestrated AI systems. The firm lists several barriers to meeting AI strategy goals:

• Data Privacy and Cybersecurity: 42% of organizations cite these as barriers to meeting AI strategy goals
• Data Quality: 34% of organizations struggle with data quality issues that impede AI deployment
• Regulatory Uncertainty: 31% of organizations face challenges from unclear regulatory frameworks
• Risk Management and Governance Gaps: 24% of organizations lack adequate governance structures

Looking ahead 12 months, respondents ranked risk management as their top challenge at 43%, followed by data quality at 36%, measurable ROI at 32%, employee adoption at 31%, and integration with other emerging technologies at 27% .

The foundation for orchestrated AI systems requires more than just deploying agents. IBM's perspective on the "agentic enterprise" emphasizes that nearly every major research source identifies data fragmentation and quality as the biggest blockers to AI value. Decades of system sprawl have left enterprises with layered data estates poorly suited for AI workflows. The shift from cloud migration to cloud intelligence requires building AI-native architectures, deploying intelligent agents across environments, managing model economics, enforcing responsible AI governance, and enabling continuous learning .

How to Build Toward Enterprise-Scale AI Orchestration

Organizations seeking to move from isolated AI pilots to coordinated systems need to address several foundational areas:

• Invest in Data Infrastructure: Establish unified data platforms and governance frameworks that enable AI agents to access and share information across workflows
• Develop Governance and Risk Frameworks: Create clear policies for AI agent behavior, decision-making authority, and escalation procedures before scaling deployment
• Build Workforce Capability: Organizations confident in their AI talent pipeline are nearly four times as likely to report meaningful business outcomes, 77% to 20%, suggesting workforce readiness is closely associated with performance
• Modernize Legacy Systems: Transition to cloud-native architectures and manage technical debt to support AI at scale while maintaining operational stability

The distinction between AI leaders and non-leaders reveals the importance of these foundational investments. AI leaders are more likely than non-leaders to prioritize revenue growth through new products, services, and AI-enabled experiences, 33% to 28%, and they place more weight on human-AI collaboration, governance, trust, and security. Non-leaders, by contrast, are more likely to prioritize cost reduction, 32% to 25%. AI leaders also report much greater confidence than non-leaders in measuring AI's effect on revenue, profitability, decision-making, and risk: 48% versus 27% on revenue, 50% versus 28% on profitability, 49% versus 32% on decision-making, and 45% versus 25% on risk .

The same pattern holds for governance and oversight. AI leaders invest more heavily in infrastructure, security, risk, and compliance, and report stronger board coverage of AI topics, 89% to 76%, deeper board-level AI expertise, 45% to 20%, and greater governance readiness, 81% to 63% .

What Does the Path Forward Look Like for Enterprise Leaders?

CIO.com research highlights that 75% of CIOs expect to deepen their involvement in AI initiatives, and IT teams are leading adoption efforts across the enterprise. However, success is not guaranteed. The research shows a clear divide between organizations realizing returns on AI and those that are not. CIO.com's latest State of the CIO research emphasizes that moving from experimentation to execution requires addressing the "production gap," the difficulty of turning promising ideas into scalable, enterprise-grade solutions .

The shift toward agentic enterprises also requires a new generation of practitioners. IBM notes that modern builders are essential because AI agents cannot deploy, govern, or scale themselves. Low-code platforms accelerate development, but disciplined engineering ensures systemic stability. Modern builders turn AI potential into real impact by designing the platforms, workflows, and guardrails that let AI agents operate reliably at scale .

Regional differences add another layer of complexity. KPMG found that the Americas lead in enterprise-scale deployment at 35%, compared with 22% in EMEA and 23% in ASPAC. The report also found different operating assumptions around human and AI control, with 41% of organizations in the Americas expecting humans to manage and direct AI agents, while 38% in ASPAC expect AI agents to take lead roles in managing projects .

The bottom line is clear: the next phase of enterprise AI success depends less on deploying more agents and more on orchestrating the ones already in place. Organizations that invest in governance, data infrastructure, workforce capability, and cross-functional coordination will emerge as the AI leaders of the next decade. Those that continue to treat AI as isolated pilots risk falling further behind.