AI Can Create Value. But Can Your Company Capture—and Keep—it?
Companies have spent the first phase of enterprise AI asking a fairly straightforward question:
Can AI make us more productive?
Increasingly, that may be the easiest question.
The harder questions come next:
- Did the productivity create economic value?
- Who captured that value?
- And can the company keep capturing it after competitors respond?
That gives us a useful way to think about the next phase of enterprise AI:
The framework sounds simple. The economics are not.
Productivity is only the beginning
A recent Harvard Business Review article, AI and the Looming Competition for Margin, makes an uncomfortable argument.
AI can improve productivity in several familiar ways: companies can produce the same output with fewer inputs, produce more with the same inputs, or create entirely new products and business models.
But if those gains spread across an industry, something else happens.
Competition responds. Lower costs can lead to lower prices. Greater output can create supply pressure. Successful innovations can be copied. What begins as a productivity advantage can become the new minimum required to remain competitive.
That creates a paradox:
AI may make companies more productive while making it harder for them to preserve margins.
The better the technology works, the stronger the competitive response may become.
That is not the AI story most companies have been preparing for.
Step 1: Create
The first question is still essential: Did AI make anything economically better?
Not:
- How many employees use the tool?
- How many prompts were submitted?
- How many tokens were consumed?
- How many agents were deployed?
Those numbers may tell us something about adoption. They do not tell us whether the business improved.
Eventually the evidence has to show up somewhere that matters:
- lower cost
- higher output
- increased revenue
- greater capacity
- faster cycle times
- better customer outcomes
- lower risk
- stronger margins
The HBR authors make a similar point when they argue that companies should measure real impact rather than convenient inputs such as token usage or lines of code.
So the first distinction is straightforward:
AI activity is not AI value.
But even that is only the first step.
Step 2: Capture
Suppose the productivity gain is real. Now ask the question that may matter even more: Who captured it?
- Maybe shareholders captured it through higher margins.
- Maybe employees gained additional capacity.
- Maybe customers captured it through lower prices or better service.
- Maybe suppliers gained bargaining power.
- Or maybe competitors adopted similar technology so quickly that the original advantage largely disappeared.
This is easy to miss because value creation and value capture are often treated as the same thing.
They are not. A company can generate real economic value and still retain surprisingly little of it.
That leads to another distinction:
Value creation ≠ value capture
And suddenly a positive productivity story becomes a strategic question.
Step 3: Defend
Now suppose the company successfully creates and captures value. There is still one more problem: Can it keep the advantage?
This may become increasingly difficult in AI because many firms can access similar foundation models, tools, and infrastructure.
The HBR authors argue that broad accessibility may make AI application advantages difficult to defend, because competitors can often work with the same underlying technologies.
That changes the meaning of AI ROI.
Imagine an initiative generates $2 million of value this year. Next year it generates another $2 million. Internally, the dashboard still looks healthy. Positive ROI.
But competitors have doubled their productivity gains, lowered prices, and improved service.
Your AI initiative still works. Your competitive position is getting worse.
That gives us another useful distinction:
Absolute AI ROI ≠ competitive AI advantage
This is where the economics become uncomfortable.
Only relative gains may matter
One of the strongest ideas in the HBR article is that efficiency gains mean less if competitors improve faster.
In a competitive market, being better than you were yesterday may not be enough. You may have to be improving faster than everyone else.
The authors put the issue plainly: only relative gains count.
That suggests a very different AI race.
The race may not be: Who adopts AI first?
It may increasingly be: Who creates value, captures enough of it, and keeps doing so as competitors respond?
AI adoption is not the race. AI value creation may only be the first lap. Value capture—and the ability to keep capturing it—is the race.
So where does the moat move?
If increasingly capable AI becomes broadly available, simply having access to intelligence becomes less differentiating.
The competitive advantage may move outward from the model. Toward:
- proprietary information
- better workflows
- stronger economics
- faster learning
- better capital allocation
- customer relationships
- distribution
- operational execution
- better decisions
And perhaps most importantly: The organization’s ability to decide, adapt, allocate, redesign, measure, and execute faster than competitors may itself become the moat.
That is a different way of thinking about AI advantage.
The model matters. But the management system around the model may matter more over time.
The questions executives should be asking next
If this argument is right, leadership needs visibility well beyond AI adoption.
Executives increasingly need to know:
- Which AI initiatives are actually creating value?
- Where is that value being captured?
- Is the advantage strengthening or deteriorating?
- Which initiatives deserve more capital?
- Which need redesign?
- Which should be routed to cheaper architectures?
- Which should stop?
And perhaps the most uncomfortable question: Are we making those decisions faster than our competitors?
These are the questions behind Create → Capture → Defend.
Over the next several articles, I’ll examine each one in more detail.
Because as AI becomes easier to deploy, the strategic challenge may move somewhere else entirely:
AI may make productivity easier to create and competitive advantage harder to keep.