Three years after generative AI triggered a new era of artificial intelligence, nearly nine out of ten organizations are using AI regularly. But McKinsey's latest Global Survey on the state of AI reveals an uncomfortable truth: Most are still stuck in pilot purgatory, experimenting without capturing meaningful enterprise-level value.
The gap between AI adoption and AI impact has never been wider—or more expensive.
88% of survey respondents report their organizations use AI in at least one business function, up from 78% a year ago. Two-thirds now use AI in multiple functions, and half report AI deployment across three or more functions.
But when you ask about enterprise-wide financial impact, the numbers collapse. Only 39% attribute any EBIT impact to AI use, and among those, most report less than 5% of their organization's EBIT is attributable to AI.
Nearly two-thirds of respondents say their organizations have not yet begun scaling AI across the enterprise. They're still experimenting or running pilots—testing AI in isolated pockets without deep integration into workflows and processes.
The survey, conducted between June and July 2025 with 1,993 participants across 105 countries, paints a picture of widespread adoption meeting stubborn growing pains. Organizations have the technology. They lack the transformation capability to extract value from it.
62% of organizations are at least experimenting with AI agents—systems based on foundation models capable of planning and executing multiple steps in workflows autonomously. 23% report scaling agentic AI somewhere in their enterprise.
But "scaling" here means limited deployment. Most organizations scaling agents are only doing so in one or two functions. In any individual business function, no more than 10% of respondents report scaling AI agents.
Agent adoption concentrates in IT and knowledge management, where use cases like service-desk automation and deep research have matured quickly. By industry, technology, media and telecommunications, and healthcare lead adoption.
The data suggests organizations recognize agents as the next frontier but remain early in figuring out how to deploy them effectively at scale.
Larger organizations lead in moving beyond pilots. Nearly half of respondents from companies with more than $5 billion in revenue have reached the scaling phase, compared with just 29% of those with less than $100 million in revenues.
The resource gap is real. Scaling AI requires investment in talent, infrastructure, change management, and workflow redesign—capabilities smaller organizations struggle to fund. The survey shows larger companies are more likely to hire for AI-related roles, with software engineers and data engineers most in demand.
But size alone doesn't determine success. The survey identifies a subset of "AI high performers"—just 6% of respondents—who report both significant value from AI use and EBIT impact of 5% or more attributable to AI. These organizations share specific practices that distinguish them from peers, regardless of company size.
AI high performers think bigger. They're more than three times more likely than others to say their organizations intend to use AI for transformative business change, not just incremental improvements.
While 80% of all respondents say efficiency is an objective of AI initiatives, high performers are significantly more likely to also set growth and innovation as objectives. They're not choosing between cost reduction and revenue growth—they're pursuing both simultaneously through AI.
This factor shows one of the strongest correlations with achieving meaningful business impact. High performers are nearly three times as likely as others to fundamentally redesign workflows when deploying AI, rather than simply automating existing processes.
In most business functions, AI high performers are at least three times more likely than peers to report scaling their use of agents rather than remaining in pilot phases.
They're three times more likely than peers to strongly agree that senior leaders demonstrate ownership of and commitment to AI initiatives. Leaders actively champion adoption, including role modeling AI use themselves.
Over one-third commit more than 20% of their digital budgets to AI technologies, compared to smaller allocations from other organizations. About three-quarters have scaled or are scaling AI, compared to one-third of others.
They're more likely to define processes for when model outputs need human validation, establish agile delivery organizations, embed AI into business processes, and track KPIs for AI solutions. All of these practices correlate positively with value capture.
While enterprise-wide EBIT impact remains rare, many organizations report benefits from individual AI use cases.
Cost benefits concentrate in:
Revenue increases appear most often in:
Qualitative benefits are more widespread:
The pattern suggests AI is delivering use-case-level value that hasn't yet translated to material enterprise financial impact for most organizations. The aggregation problem—turning scattered wins into systematic advantage—remains unsolved.
Respondents show divergent expectations about AI's effect on workforce size over the next year:
Looking at individual business functions, a median of 17% report workforce declines in the past year due to AI, but a median of 30% expect decreases in the year ahead—suggesting anticipated acceleration in AI-driven displacement.
Larger organizations are more likely than smaller ones to expect enterprise-wide workforce reductions. AI high performers are more likely than others to expect meaningful change in either direction—either workforce reductions or increases.
Simultaneously, most respondents—especially from larger companies—report their organizations hired for AI-related roles over the past year. The talent needs focus on software engineers and data engineers.
The paradox: Organizations are hiring AI specialists while anticipating AI will reduce overall headcount. The transformation creates new roles while eliminating others, with net effects varying by organization.
51% of AI-using organizations have experienced at least one negative consequence from AI deployment. Nearly one-third report consequences stemming from AI inaccuracy—the most commonly cited risk.
Organizations are taking risk mitigation more seriously than in previous years. Respondents report acting to manage an average of four AI-related risks today, compared to just two in 2022.
The risks organizations work to mitigate largely align with risks they've experienced consequences from. However, explainability—the second-most-commonly-reported risk—is not among the most commonly mitigated, suggesting a gap between acknowledged concerns and protective action.
AI high performers, who have deployed twice as many use cases as others, are more likely to report negative consequences—particularly related to intellectual property infringement and regulatory compliance. They also attempt to protect against a larger number of risks, possibly because broader deployment exposes them to more risk categories.
McKinsey's data reveals what separates AI experimentation from AI value: transformation mindset and execution capability.
Organizations capturing significant value aren't just deploying AI tools—they're redesigning how work gets done, setting ambitious objectives beyond cost reduction, securing committed leadership, investing substantially, and implementing disciplined management practices.
The pilot-to-scale transition remains the critical bottleneck. Organizations can adopt AI easily. Integrating it deeply enough to generate enterprise-level returns requires organizational transformation that most have not yet undertaken.
The survey suggests a bifurcation emerging in the market. A small group of high performers—6% of respondents—are pulling away, capturing disproportionate value through systematic approaches to AI deployment. The remaining 94% are using AI but not yet transforming with it.
For organizational leaders, the McKinsey findings offer both warning and roadmap.
The warning: AI adoption without transformation capability produces expenses without corresponding returns. Experimenting indefinitely in pilot phases burns resources and time while competitors who crack the scaling challenge build durable advantages.
The roadmap: The practices distinguishing high performers are knowable and implementable. Ambitious objectives, workflow redesign, committed leadership, substantial investment, and disciplined management practices collectively enable value capture at enterprise scale.
The window for catching up may be narrowing. As high performers deploy agents at scale, redesign workflows systematically, and compound their advantages quarter over quarter, the gap between leaders and laggards widens.
Three years into the generative AI era, adoption is table stakes. Transformation is the differentiator. Most organizations have cleared the first hurdle. The second one—where value actually materializes—is proving far more difficult to cross.
If your organization is navigating the transition from AI pilots to scaled deployment and needs strategic guidance on transformation practices, workflow redesign, and value capture, Winsome Marketing's team can help you implement the approaches that distinguish high performers.