AI adoption is nearly universal. Competitive advantage from AI is not. The reason has nothing to do with the technology - and everything to do with what happens when every competitor buys the same tool from the same vendor.
Somewhere in mid-2025, the question in every strategy meeting shifted. It stopped being "are you using AI?" and became "which AI are you using?" The shift felt like progress. It wasn't.
I've sat through enough quarterly reviews and vendor briefings over the past eighteen months to recognize the pattern. A company announces its AI initiative. A platform gets deployed. Productivity metrics move. A leadership team reports back that adoption is on track. And then, six months later, twelve months later, nothing meaningful has changed. Revenue is flat. Competitive position looks the same as it did before the rollout. The company has AI - and no advantage.
McKinsey's April 2026 analysis, "Where AI Will Create Value - and Where It Won't", put a number on something practitioners have been watching happen in real time: nine out of ten companies have deployed AI, but 94% report not seeing significant business value. The report's sharpest observation is also its most uncomfortable one - that productivity improvement, on its own, is "unlikely to expand profit pools or provide a durable advantage." Not because AI doesn't work. Because when everyone has it, the gains cancel out.
This is the structural problem that most AI strategy discussions skip entirely.
When Everyone Buys the Same Tool, the Advantage Is Zero
The logic of competitive advantage has always been straightforward: do something better than your competitor, and you can charge more, grow faster, or retain customers longer. The assumption underneath that logic is that doing it better requires something your competitor doesn't have.
AI is now available to every company, at scale, from the same three or four major vendors, on the same pricing tiers. The executive who approved your AI contract almost certainly approved the same contract at a competitor twelve months earlier or later. You're running the same models. You're accessing the same underlying capabilities. The efficiency gains are real - but they're real for everyone.
Think about what that means structurally. If your entire industry reduces its cost per support interaction by 30% because every player deploys the same AI contact center tooling, what has actually changed? Prices fall. Margins normalize. The productivity gain flows through to customers or gets competed away in pricing. The cost structure of the industry shifts, but no individual company ends up ahead.
This is not a hypothetical. I watched a version of it play out during my years at Zendesk, as wave after wave of productivity tooling became table stakes across enterprise customer service. Ticketing automation, self-service deflection, SLA routing - each one looked like a differentiator for the first movers and became a baseline expectation within two years. AI is compressing that cycle from years to months.
The Mistake Hiding Inside the AI Business Case
Most AI deployments inside enterprises are justified on cost. The business case is efficiency: reduce headcount, deflect volume, cut cost per transaction. Those are real and legitimate outcomes. But cost reduction is not competitive advantage. It is necessary to compete - but not sufficient to win.
The companies I've seen actually pull ahead - at Adobe, at the Intelegencia clients we work with across enterprise and PE-backed businesses - aren't the ones deploying AI to do the same things cheaper. They're using it to do things that weren't previously possible at scale: encoding institutional knowledge that used to live in one person's head, systematizing judgment calls that took years of experience to make reliably, running personalization at a depth that no competitor can replicate without the same proprietary data infrastructure.
That distinction matters more than anything in the AI business case. Efficiency from AI is available to everyone. Capability from AI - the kind built on data moats, process architecture, and organizational knowledge that took years to accumulate - is not.
The uncomfortable question for any enterprise AI strategy is: what does our deployment actually do that a competitor with the same budget couldn't replicate in six months? If the honest answer is nothing, then you're building operational competence, not competitive advantage.
What Actually Creates Durable Differentiation
The companies generating real, compounding advantage from AI share a structural characteristic: they're using AI to operationalize something that was previously locked inside people, relationships, or institutional memory - and that is genuinely hard to replicate.
This looks different depending on the business. In customer success, it can mean building AI-assisted health scoring on top of fifteen years of retention data that no competitor has. In sales, it means using AI to encode the nuanced territory knowledge that typically walks out the door when your top rep leaves. In operations, it means systematizing workflow decisions that previously required a senior person making judgment calls - and making those calls consistently, at scale, without the senior person in the loop.
What these have in common isn't the AI technology. It's what the AI is being applied to. Proprietary data. Hard-won processes. Organizational depth that took time to build and can't be purchased from a vendor.
There's a parallel here to how the best enterprise software companies built moats in the 2010s. The software itself wasn't the barrier - the network effects, the data integrations, the switching costs were. The same logic is starting to apply to AI. The model is the commodity. The context you train it on, the process it's embedded in, the organizational muscle built around it - that's where the advantage lives.
The Operational Reality Most Leaders Aren't Discussing
Here's what I don't hear in enough AI strategy conversations: the gap between deploying AI and generating advantage from it is an organizational problem, not a technology problem.
The companies I've seen struggle aren't failing because the AI doesn't work. They're failing because the AI is deployed on top of fragmented data, inconsistent processes, and an org structure that was designed before any of this existed. The AI surface is clean. Everything underneath it is not. And no model compensates for that.
I worked through versions of this with organizations scaling from several hundred to several thousand customers - situations where the operational infrastructure was built for one stage of growth and then held together with workarounds at the next. Deploying AI into that environment doesn't fix the fragmentation. It accelerates it. The model surfaces the inconsistency faster, at higher volume, across more touchpoints simultaneously.
The hardest conversations I've had in this space aren't about which model to deploy. They're about whether the underlying data is coherent enough to give the model something meaningful to work with. Most of the time, it isn't. And closing that gap takes months, sometimes years - which is exactly why companies that start early with the data infrastructure work tend to pull ahead, while late movers find the tooling available but the foundation not ready.
Where the Real Advantage Window Still Exists
This isn't an argument against AI investment. It's an argument for being precise about what the investment is actually buying.
Efficiency from AI is real, available now, and worth pursuing - but it is table stakes. Every competitor is doing it. The companies that will look back in 2028 and say AI genuinely changed their competitive position are the ones building toward something that can't be replicated by signing the same enterprise contract.
That means different things at different maturity stages:
- For companies in early AI deployment:
- Don't treat efficiency as the endpoint - treat it as the first phase
- Begin mapping where your proprietary data, institutional knowledge, and differentiated processes actually live
- Identify what your org can do at AI-assisted scale that a competitor starting from scratch could not easily replicate
- For companies with AI already deployed:
- Audit honestly: are you using AI to be cheaper, or to be different?
- Identify the one or two areas where your data, your client relationships, or your process depth create genuine asymmetry - and concentrate AI investment there
- Stop measuring AI success by adoption rate and start measuring it by capability gap
The efficiency wave from AI is real and it's moving through every industry. The question is whether you're riding it with everyone else, or using it to build something they can't easily follow.
