Most Companies Still Stuck in AI Pilot Phase—Missed EBIT Impact Creates Rerating Risk

The journey from AI pilot to enterprise-scale impact follows a familiar historical arc, marked by widespread initial enthusiasm and a persistent gap between promise and measurable results. Today, the pattern is clear: most companies are still in the early stages of scaling, mirroring the uneven adoption seen during the enterprise resource planning (ERP) wave of the 1990s.

The critical metric is the sheer number of organizations that have not yet begun scaling AI across the enterprise. Nearly two-thirds of respondents in the latest McKinsey survey say their companies are still in the experimentation or piloting phase, with only about one-third having begun scaling programs. This lag is the defining feature of the current adoption curve. It shows that while AI tools are now commonplace, embedding them deeply enough into workflows to realize material enterprise-level benefits remains a work in progress.

This sets up a classic productivity paradox. On one hand, the reported enablement of innovation is high, with 64% of organizations saying AI is enabling their innovation. On the other, measurable impact on earnings before interest and taxes (EBIT) at the enterprise level is much lower, reported by just 39%. This gap between perceived potential and actual financial impact echoes the early cloud computing era, where initial efficiency gains were followed by a longer, more complex period of realizing transformative business value. The early cloud adoption curve also saw a similar lag between technical deployment and the redesign of business processes needed for full payoff.

The acceleration of worker access, however, is changing the dynamics. Worker access to AI rose by 50% in 2025, creating a surge in demand for practical application. This expansion from tool access to effective application is the new imperative. The data shows that only 34% of organizations are truly reimagining their businesses, while another third are merely redesigning key processes. The rest are using AI at a surface level. The historical lesson is that scaling success comes not from simply deploying more tools, but from the deliberate, top-down effort to redesign workflows and business models around the new capabilities-a shift that most companies have yet to make.

The Fluency Imperative: Skills as the New Bottleneck

The expansion of worker access is creating a new bottleneck: the workforce's ability to use AI effectively. This is the modern equivalent of the digital literacy gap that stalled productivity gains during the early internet boom. True AI fluency means understanding the technology's strengths and limitations, and applying it productively to real problems-a capability that is quickly becoming table stakes across industries.

The skills gap is the top barrier to integration, and education is the primary talent strategy companies are deploying. This mirrors the early resistance to workflow changes seen during the enterprise resource planning (ERP) wave, where the focus was often on training people to use new software rather than redesigning business processes around it. The latest data shows that education-not role or workflow redesign-was the No. 1 way companies adjusted their talent strategies due to AI. This suggests a common pattern: organizations are investing in upskilling to keep pace with the technology, but not necessarily in rethinking how work gets done.

The contrast between high-performing companies and the rest is stark. The most successful adopters are not just using AI for efficiency; they are using it to drive growth and innovation. According to the survey, the companies seeing the most value from AI often set growth or innovation as additional objectives. This is the critical difference. High performers are actively redesigning workflows, with half of those AI high performers intending to use AI to transform their businesses. This workflow redesign is the proven path to transformative impact, a pattern seen in successful cloud migrations where the real payoff came from process reengineering, not just infrastructure shifts.

The bottom line is that measuring AI fluency is not a vanity exercise. It is essential for identifying the gaps between tool access and effective application. Without this fluency, teams risk working harder, not smarter-spending hours fixing AI-generated errors or avoiding useful capabilities. In the end, the historical lesson is clear: scaling success comes not from deploying more tools, but from building the human capability to use them wisely.

Metrics and Mechanisms: Bridging the Gap

The path from AI tool access to tangible P&L impact is paved with operational changes, not just technology deployment. The core risk is that teams using AI without true fluency often work harder, not smarter. This is a familiar pattern from poorly implemented legacy systems, where new software created more friction than efficiency. Without understanding the technology's strengths and limitations, employees may spend hours fixing AI-generated errors or second-guessing recommendations, leading to wasted effort and flawed decisions. In this scenario, the AI investment becomes shelfware, its potential value unrealized.

The proven mechanism to overcome this risk is workflow redesign. This is the critical success factor that separates high performers from the rest. The data shows that half of those AI high performers intend to use AI to transform their businesses, and most are actively redesigning workflows. This mirrors the re-engineering required for past technology rollouts, like the enterprise resource planning (ERP) systems of the 1990s. The payoff came not from automating old processes, but from fundamentally redesigning them around the new capabilities. The same principle applies to AI: embedding it deeply enough to drive enterprise-level EBIT impact requires a deliberate shift in how work gets done.

Measuring AI fluency is the essential feedback loop that drives this process. It moves the conversation from vague promises to quantifiable outcomes. By tracking usage and proficiency, organizations create a data-driven system for continuous improvement. This measurement identifies gaps-like departments lagging in adoption-and guides targeted training. More importantly, it quantifies impact: if an AI tool is saving 10% off project times, that's a clear ROI signal to double down. If not, it's a prompt to refine the approach. As the lesson from past rollouts teaches, you can't improve what you don't measure. In the end, the goal is to turn AI from a costly experiment into a lever for sustained, measurable business value.

Catalysts and Risks: The Path to Scale

The forward path is now defined by a stark urgency. The catalyst for action is clear: the number of companies with ≥40% of AI projects in production is set to double in six months. This rapid acceleration from pilot to scale creates a narrow window for organizations to close the fluency gap. Those that fail to build workforce capability in time risk falling behind in the race to operationalize AI.

Yet the most significant barrier may not be technology, but expectation. A deep chasm exists between leadership and the front line. According to a recent survey, 78% of leaders believe their organizations have a clear, long-term AI strategy, while only 39% of employees agree on coordinated implementation. This disconnect is a primary risk. It breeds confusion, undermines buy-in, and sets up projects for failure. When employees don't see a coherent plan, they are more likely to avoid new tools or use them incorrectly.

The ultimate risk of inaction is that AI investments become expensive shelfware. Despite the promise of efficiency, the tools sit unused or misapplied, failing to generate the promised productivity gains. This is the modern version of the legacy system that created more friction than efficiency. Without AI literacy across the organization, teams may spend hours fixing flawed outputs or second-guessing recommendations, leading to wasted effort. As one expert notes, your expensive AI investments gather dust while employees stick to familiar spreadsheets and manual processes.

The path to scale, therefore, hinges on bridging this gap between ambition and activation. It requires leaders to move beyond strategy statements and actively communicate, align, and train. The historical lesson is that technology rollouts succeed when they are matched by a parallel effort to change how work gets done. The catalyst is here, but the window to act is closing.