Picture a CEO at a regional company that has had a good year. Margins are steady, the best customers renewed, and the board meeting was short. Over coffee afterward, a director asks a simple question: of all the ways a competitor could use AI against you, which one frightens you most?
Most leaders answer quickly, and the speed of the answer is the interesting part. This article is about why that answer is rarely as well considered as it feels, what it reveals about the map you are navigating by, and why this particular moment is a poor one to leave the question unasked.
A Spectrum, Not a Switch
Every organization splits its attention between two jobs. One is stewardship: protecting, running, and steadily improving what already works. The other is exploration: searching for new ways to create value, knowing most searches will not pay. Researchers have studied this tension for decades. James March’s 1991 paper in Organization Science argued that learning systems which refine current practice faster than they search for alternatives tend to succeed in the short run and erode over time. In 2004, Charles O’Reilly and Michael Tushman reported in the Harvard Business Review that firms doing both well tended to keep exploratory units separate from traditional ones while a single senior team coordinated them.
Those papers describe a tradeoff, not a verdict. We find it useful to treat it as a spectrum with stewardship at one end and exploration at the other, and to notice that your competitors are not all standing at the same point on it. The names and positions below are our own shorthand, not an established taxonomy, and the competitors are composites, not real companies.
Five Competitors, Five Points on the Spectrum
The first is the Steady Steward. This rival runs the same model with the same map and does it well, and treats AI as a tool to adopt slowly and carefully. In the short run this competitor is the least frightening, because it is playing the game you already know. It becomes dangerous only if its map matches the territory better than yours does.
The second is the Disciplined Adopter. This rival is still a steward, but it applies AI hard to cost, speed, and quality inside its existing model. You will meet this competitor in pricing and in service levels, not in headlines. It is the one most likely to feel frightening in the next two years, because the damage shows up in margin. In the language of our four shapes of the AI dividend, this is the Slow Melt: not a sudden loss, but a steady one.
The third is the Ambidextrous Rival. This competitor protects its core and funds a separate team that explores, much as O’Reilly and Tushman described. Today it looks unremarkable. Its strength is that it holds options, so when the territory shifts it is already positioned to move. It is rarely the loudest threat, and it may be the one that matters most over five years.
The fourth is the Reinventor. This rival rebuilds its offering around what AI makes possible and accepts that it will cannibalize its own revenue to do it. Picture a competitor that reprices a service from hours to fixed bids because AI changed its cost to deliver. You will feel it in deals you thought were yours. Whether it is existential depends on your market; where one or two providers can come to dominate because delivery scales like software, the Reinventor is the competitor to fear.
The fifth is the Newcomer With No Map. This is an AI-native startup or an investor-backed roll-up with no legacy systems, no history of what has failed before, and no assumptions about what your market will bear. It is the hardest to see, because it does not play your game, and the hardest to predict, because it has nothing to protect.
How to Score Them
Give each of the five two scores from one to five. The first measures how frightening it is: how likely it is to reach your customers in the next two years. The second measures how dangerous it is: if it works, could your company survive it as it is today? Then name a real competitor, or a real possibility, for each slot. Do it alone first, then ask two colleagues to do it without seeing your answers, and compare.
Here is our reading of what the scores tend to show, and it is a judgment and not a finding. Leaders usually fear the competitor closest to their own position on the spectrum, because it is the one they can imagine. A steward fears the Disciplined Adopter and pays little attention to the Reinventor. An explorer fears the Steady Steward’s cost advantage and overlooks the quiet one with options. The competitor you fear most is often a portrait of your own assumptions. If every score for the far end of the spectrum is a one, that is worth a second look.
What the Surveys Can and Cannot Tell You
Two large surveys suggest that leaders expect change faster than they have realized it. Gartner’s CEO and senior executive survey, covering 469 respondents with fieldwork ending in the fourth quarter of 2025, found that 80 percent expect AI to force a high or medium degree of change to their operational capabilities. Fifty-four percent said their automation is limited to specific tasks today, and only 13 percent expect to remain at that level by the end of 2028. PwC’s 2026 Global CEO Survey, covering 4,454 CEOs in 95 countries and territories, fielded from late September to early November 2025, found that 12 percent said AI had improved both costs and revenue, 33 percent reported a gain in one of the two, and 56 percent reported no significant financial benefit so far.
Treat these with care. They are self-reported by executives, they were collected before this fall, and neither measures how well anyone’s map matches reality or what any competitor is doing. What they are consistent with, and this is our inference and not their conclusion, is a situation where many leaders sense that the ground is moving and few have yet turned that sense into a changed strategy.
Stewardship Is Not the Villain
None of this is an argument against stewardship. Most companies exist because someone stewarded something valuable for a long time: customer trust, a regulated license, a reliable operation, a capital base. When the map a leader is navigating by matches the territory, stewardship is the right strategy, and exploration beyond a modest allowance is waste. A steward with an accurate map should not be talked out of anything.
The condition is the word accurate. A map is a set of assumptions about what things cost, which capabilities are scarce, what customers will pay for, and what is possible. We wrote about how those maps go stale quietly in The AI Earthquake · Why Your Old Map Doesn’t Match the Territory Any More, and about how comfort, more often than disaster, ends expeditions in our piece on the lotus-eaters. Our judgment is that in the earliest stage of the AI revolution very few maps match the territory, because the tools, their prices, and what they can do are changing faster than any budget cycle revises its assumptions.
You can test your own map in an afternoon. Ask when you last checked your assumptions about what is hard, slow, or expensive against what AI can do today at low cost. Ask which of the decisions your team remembers as “we tried that and it did not work” were made before 2025. And ask what a competitor with no history in your market would assume about it. If those questions produce uncomfortable answers, the map is telling you something.
What This Will Not Do
A scorecard will not predict your competitors. The five types are simplifications, and a real rival can sit at two points at once or move between them. Transformation also carries real risk: exploration spends money, most experiments fail, and a company can lose its core while chasing the edge. Our argument is not that every company should reinvent itself. It is that the decision should be made on a map you have tested, not on one you have inherited. If you test it and it holds, steward on, and revisit in a year.
The Case for Transformation, Now
Our view, and admittedly it’s a biased view – not a measurement – is this: AI is changing faster and more powerfully than any technology in history. It is a commoditization of cognition, and we believe both its potential and its risks exceed those of any economic revolution humankind has traversed. We cannot prove that comparison, and no one can yet, but we would rather you weigh the argument than wait for the proof.
A practical first step is small. Run the scoring exercise with your leadership team this month, test your map against the three questions, and decide where you sit on the spectrum on purpose and not by default. If you want an outside view, our independent assessment gives you the state of your technology and AI in your own terms, and our bench of more than 450 former CIOs and CTOs includes experts who have led companies like yours through exactly this kind of choice.
If you are not going to consider a strategic transformation in this moment, one wonders what combination of circumstances might ever warrant it.
