Everyone Has Adopted AI. Almost Nobody Changed.
Blog by Clay Wolfe
There's a moment in most of my client conversation now where someone says, "Oh, Yes, we're using AI."
Then you dig deeper to see how it has transformed their business and there is a brief pause. Eventually they mention that the marketing team gets through first drafts faster or that they believe IT is using it create efficiencies.
That pause is the most important data point in enterprise AI right now.
The numbers agree. McKinsey's latest global survey found 88% of organizations regularly use AI in at least one business function, essentially universal adoption. But nearly two-thirds haven't begun scaling it across the enterprise, and only 39% report EBIT impact at the enterprise level. MIT's widely-cited NANDA study is more blunt: 95% of enterprise GenAI pilots produced no measurable P&L return. The methodology of that study took some fair criticism, but nobody has produced a more flattering number since.
Buying AI turns out to be the easy part. Owning the new fancy shoes does not make you a runner.
Adoption is not transformation
In CMG's AI Disruption work, we look at organizations across three dimensions: Offering (what you sell), Operations (how you run), and Engagement (how you connect), while considering AI Maturity across all three.
The pattern is remarkably consistent. Most organizations are deploying AI into Operations, in pockets, and calling it a strategy.
Meanwhile, AI disruption is occurring across all three of these dimension.
That's the uncomfortable part. AI's first casualty isn't your cost structure it's your value proposition. Associations built durable businesses on information access; AI hands it out for free. SaaS companies priced per seat; AI agents don't need seats. If the core of what you sell is something a model now produces at near-zero marginal cost, a faster back office is not going to save you.
The market has started pricing this
For our private equity clients, this stopped being a philosophical conversation a while ago.
EY's 2026 Global PE Exit Readiness study found that buyers now explicitly distinguish between AI activity and AI strategy. A list of pilots and productivity licenses doesn't support a valuation. What supports a valuation is AI embedded in the operating model, benefits that are actually measured, governance that exists on paper and in practice, and data clean enough to provide evidence to the claim under diligence.
AI readiness is becoming part of the equity story. "We have enterprise ChatGPT" is not an equity story.
Three questions worth more than any tool evaluation
1. Outside in. What do we charge for that AI now gives away? Answer it honestly, then rebuild around what's left: community, credentialing, outcomes, judgment, trust. This one is true whether or not you ever deploy a thing.
2. Inside out. Have we redesigned any work, or just accelerated it? McKinsey's high performers are roughly three times more likely to have fundamentally redesigned workflows. A faster bad process is still a bad process.
3. Proof. If we can't show the numbers, we can't claim the win. Not a slide someone built after the fact, a baseline from before, and a number an outsider could reconstruct.
None of this requires a moonshot. It requires starting from the business problem rather than the vendor demo, which is less exciting, but roughly 95% more likely to work.
The organizations landing on the right side of disruption aren't the ones running the most pilots. They're the ones willing to ask whether the thing they sell still holds.
