Three (3) years into the generative AI era, the headline statistic should disturb every C-suite leader: BCG found that 74% of companies have yet to show any tangible value from AI despite widespread investment. McKinsey puts the same finding another way: 88% of organizations use AI in at least one function, but only 39% report any EBIT impact at the enterprise level. This blog dissects the structural reason why, and what it takes to cross from spending to returning.
Three (3) years into the generative AI era, most C-suite leaders share a version of the same private frustration. The investment is real. The announcements were made. The pilots ran. And the financial results are underwhelming.
The data confirms that this is not an isolated experience.
BCG's research found that 74% of companies have yet to show any tangible value from AI despite widespread investment. Only 4% have achieved what BCG classifies as cutting-edge AI capabilities across the enterprise. McKinsey's State of AI survey, conducted across 1,993 participants in 105 countries in 2025, puts the same finding in sharper relief: 88% of organizations use AI in at least one business function, but only 39% report any EBIT impact at the enterprise level. Of those, most see less than 5% improvement.
These two research bodies are measuring different things but arriving at the same conclusion. There is a widening gap between AI activity and AI value, and it is structural, not motivational.
Understanding that distinction is the difference between doubling down on the wrong strategy and making the one change that moves the needle.
The standard enterprise AI narrative in most boardrooms runs something like this: "We've deployed AI across multiple functions, we have active pilots, our teams are trained, and we're seeing efficiency gains in targeted areas."
All of that can be true. And the enterprise can still be in the 74%.
Here's why: use-case-level efficiency and enterprise-level financial impact are two fundamentally different measurements. McKinsey's data makes this distinction explicit. 64% of organizations say AI is enabling their innovation at the use-case level. But that use-case value is not translating upward. Only 39% see any EBIT movement.
The gap between those two numbers, 64% and 39%, is not a measurement lag. It is the architecture gap.
When AI is deployed by function, marketing buys one tool, operations buys another, and finance uses a third. Each creates isolated value within its own lane. The marketing team saves time on content generation. The operations team improves forecasting accuracy. Finance automates reporting.
None of these gains connect. They don't inform each other's decisions, share data models, or produce insights that cross organizational boundaries. Leadership still makes strategic decisions based on fragmented information, some AI-assisted and some not, with no unified intelligence layer to synthesize it.
This is the architecture problem that explains both BCG's 74% and McKinsey's 39%.
A Harvard Business Review analysis of enterprise AI deployments documented the downstream consequence: organizations running separate AI models by department reach contradictory conclusions about the same business reality. In one documented case, a risk team flagged customers as too high-risk at the same time a marketing team targeted those identical customers for growth, because each team's AI was operating on separate data with no shared intelligence layer to identify the conflict.
That is not an AI problem. It is an orchestration problem.
BCG and McKinsey approach AI performance measurement from different angles, but their findings converge in a way that every C-suite leader should understand.
McKinsey focuses on operational maturity: how widely AI is deployed, at what stage of scaling, and what functional impact is being captured. Their finding that 88% of organizations use AI but only 39% see EBIT impact is a measurement of the adoption-to-outcome conversion rate.
BCG focuses on strategic value, specifically who is capturing the gains and at what scale. Their finding that 74% have yet to generate tangible value, with only 4% achieving cutting-edge enterprise-wide capability, describes the distribution of AI returns: a winner-take-most dynamic where the top tier pulls ahead while the majority circles the same ground.
Read together, the picture is clear. Broad, function-level AI adoption is producing use-case gains for most organizations, but enterprise-level financial returns are concentrating in a small group of companies that have made a different structural choice.
What separates them is not the sophistication of their AI models. It is the presence or absence of an intelligence architecture that connects those models to strategy, to leadership decision-making, and to business outcomes.
S&P Global's 2025 analysis introduced a data point that deserves direct attention from every CIO and CEO allocating AI budget: the share of companies abandoning most of their AI projects jumped to 42% in 2025, up from just 17% the prior year.
This acceleration is significant. It indicates that the initial wave of AI enthusiasm, funded by innovation budgets and fueled by competitive pressure, is hitting a wall of unmet expectations. Organizations that cannot demonstrate measurable value are withdrawing rather than doubling down.
The underlying drivers, per multiple research reports, are consistent: cost escalation, unclear value measurement, and AI initiatives that are disconnected from core business objectives. These are symptoms of the same architecture gap, not the absence of AI capability, but the absence of the structure needed to translate capability into outcomes.
The CFO who cuts an AI program that showed no return is not making the wrong call. The problem is that the program was structured in a way that made return nearly impossible to demonstrate, because it was never connected to a measurable enterprise outcome in the first place.
BCG's research on the small group of companies achieving cutting-edge, enterprise-wide AI impact identifies a consistent set of differentiators. They are not running more pilots. They are not spending more on models. They are doing something structurally different.
Moving from the 74% to the 4% is not a technology decision. It is a leadership decision, made across four dimensions.
What are we measuring? Define enterprise-level AI success in financial terms before deploying. Use-case efficiency is a leading indicator, not the destination.
Who owns the outcome? Establish cross-functional ownership of AI returns, with business leaders and technology leaders jointly accountable, not IT alone. Deloitte's 2025 research found that when the CTO, CFO, and Chief Strategy Officer jointly own technology investment decisions, organizations are significantly more likely to see above-average EBITDA.
How are our AI investments connected? Map your current AI deployments against your strategic decisions. If your AI tools are not informing each other, they are not informing your strategy.
Where is the intelligence layer? Identify where in your architecture the synthesis happens, where your AI outputs become connected insights, accessible to leadership and trackable against business outcomes. If you cannot answer this, you have the architecture gap.
The enterprises generating real financial returns from AI are not running better point solutions. They have built, or deployed, the orchestration layer that connects their AI investments to each other, to their data, and to the decisions their leadership makes every day.
This is what shifts AI from a collection of functional experiments into a strategic asset. It is what allows leadership to ask a question about customer risk, market opportunity, or operational performance and receive an answer synthesized across every AI-enabled system in the enterprise, not fragmented across siloed dashboards.
The 74% statistic will not improve by deploying more AI tools. It improves when enterprises build the intelligence infrastructure that makes the tools they already have actually work together.
Aevah is the Enterprise Intelligence OS that connects your people, processes, and platforms, awakening the organizational intelligence already inside your systems and delivering the strategic clarity to act on it.
If you'd like to understand where your organization sits on the AI maturity curve and what it would take to cross into the top tier of enterprise AI performance, request a 20-minute AI Maturity Assessment.