PwC’s 2026 AI Performance Study found that 74% of AI’s economic value is captured by just 20% of organizations. The remaining 80% are splitting what’s left, and the gap is widening, not closing.
Most people read a stat like that and assume the losing 80% have a data problem or a governance problem. Often, they do. We’ve written before about why good data still isn’t enough and why most AI initiatives aren’t ready to scale. That diagnosis hasn’t changed. But it answers a different question than the one this study raises. Readiness explains why an initiative stalls or breaks in production. It doesn’t explain why two organizations that both clear that bar still end up years apart in value captured. For that, you have to look at what they pointed AI at in the first place.
Leaders Chase Growth. Everyone Else Chases Savings.
Here’s a divide we see, quarter after quarter, on AI roadmaps. The organizations winning are using AI to open new revenue, not just to trim cost from existing work. PwC’s data backs this up directionally; the leading cohort was far more likely to use AI for new growth opportunities and to say it changed how they thought about the business model itself. We’d add the part the data can’t capture: that shift only happens when someone in the room is explicitly responsible for asking “what could we now offer that we couldn’t before,” instead of defaulting to the easier question of what to automate.
The easier question is why most organizations are stuck. Point AI at the cost line, automate a workflow, count the hours saved – that playbook isn’t wrong. It’s just small. Cost-cutting use cases have a ceiling, because there’s a finite amount of existing work to automate. Growth use cases don’t, because they aren’t bounded by what the organization already does.
This is the distinction we push clients to sit with before they greenlight the next AI initiative. Not “what can we automate” but “what can we now offer, reach, or build that we couldn’t before.” Those are different planning conversations, and most AI roadmaps we review are still only having the first one.
Growth Ambition Without a Foundation Is Just a Faster Way to Be Wrong
None of this replaces the readiness question. AI without data discipline creates noise, not value. That’s true whether the use case is cost-cutting or growth, and it’s the first lens we apply on every engagement. What changes is where that question sits in the conversation. It’s the gate, not the whole strategy.
We see this play out as two failure modes, and they look nothing alike from the outside. An organization that fixes its data and governance but keeps pointing AI at the same cost-cutting use cases ends up reliable and small: clean dashboards reporting on a business that hasn’t moved. An organization that chases growth without the operating model to back it ends up fast and wrong – funded pilots that can’t survive contact with a real decision, because nobody made the call on who owns the outcome or what “done” means. Governed data and a working operating model aren’t competing priorities. They’re the same prerequisite, viewed from two angles.
Three Questions Before the Next AI Roadmap Review
These are the questions we want answered in the room, not after the fact, when a client brings us in to evaluate what’s on their roadmap.
1. Where in the business would new revenue or a new market actually show up if this worked the way you’re hoping?
If the honest answer is “it wouldn’t, this just makes the current process cheaper,” that’s a cost initiative wearing an AI label. Evaluate it against that ceiling, not against growth numbers.
2. Who is accountable for turning this from a pilot into a funded initiative?
Growth use cases die in the gap between “interesting demo” and “someone owns the business case.” That gap is organizational, not technical. It’s an Operating Model problem, not a tooling one.
3. If this succeeded beyond expectations, does the business have a plan to capture that upside, or would it just generate a good slide for next year’s strategy meeting?
We’ve sat in enough of those meetings to know which answer most roadmaps are actually prepared to give.
The Gap Isn’t Closing on Its Own
Without a change in approach, the distance between the leading 20% and everyone else gets wider, because the leaders keep learning faster and scaling what already works. That’s not a one-time advantage, it compounds.
We don’t think the organizations that close that gap will do it by copying the leaders’ tool stack. They’ll do it by asking what they’re actually trying to build with AI and being honest about whether “build” is even the right verb for what’s currently on the roadmap.
By Jeremy Stierwalt, Chief Data & AI Officer