On a quiet stretch of Scottish moorland, two red deer stags face off across fifty yards of open ground. It’s the rut, the one part of the year when a stag’s entire reproductive future depends on beating his rivals, and by every logical measure they should be fighting. Antlers were practically built for this. Instead, they roar. For minutes at a time, they trade bellows across the glen, pacing back and forth, sizing each other up. Only if neither backs down do they walk toward each other, parallel, shoulder to shoulder, for one more round of assessment at close range. Most of the time, that’s where it ends. One stag breaks off. The other keeps the ground. Nobody bleeds.

Nobody negotiated this. It’s just math: a real fight is expensive even for the stag who wins it, so evolution favors animals that spend most of their energy signaling and save the actual violence for the rare case the signals don’t settle things.

Once you notice this pattern in animals, you start seeing it everywhere in markets, too. Coca-Cola and PepsiCo have split roughly three-quarters of the U.S. carbonated soft drink market for decades. They compete hard: new products, ad campaigns, exclusive placement deals. But there’s one lever they leave alone, the one that would actually hurt them both, and that’s price. Microsoft, deep in a Justice Department antitrust probe over Windows in 1997, wrote Apple a $150 million check to keep it alive rather than let it disappear. American Airlines doesn’t go to war with Delta in Atlanta, because Delta could hit back in Dallas.

None of this is collusion, and none of it is charity. It’s the same cost-benefit logic that keeps the stags roaring instead of goring each other, just running on quarterly earnings instead of testosterone. And once you can name the specific mechanism behind a given rivalry, you get a sharper question to ask than “how competitive is this market?” So here’s the real question: is this restraint structural, or is it just today’s math, waiting for one input to change?

There are three specific mechanisms behind most of what looks like restraint in a concentrated industry, and they’re worth treating as three separate tools, not one blurry instinct. I’ll lay out all three here:

1. Survival — will the leader let the runner-up live?

2. Price — will rivals fight over the one number that would hurt them both?

3. Territory — will rivals invade each other’s strongest ground?

Each one shows up independently in nature and in markets, and each one should change something specific about how you analyze a company.

I. Existential Forbearance: Why the Snake Doesn’t Bite

Male rattlesnakes carry a rival’s most efficient weapon built directly into their mouths. Yet when two of them contest the same territory or the same female, they almost never use it on each other. Instead, they rear up, twist around one another, and wrestle, pushing, pinning, trying to force the opponent’s head to the ground, for as long as twenty minutes, without a single bite. Biologists have documented the same pattern of restraint across an enormous range of species. Mule deer crash antlers together but avoid an opponent’s unprotected side. Across the animal kingdom, most contests settle through display and assessment rather than a fight to the finish.

The explanation dates to 1973, when biologist John Maynard Smith showed mathematically why this makes sense. If every animal fought all-out every time, the species as a whole would do worse. Even the strongest individuals would eventually take a fight-ending injury from someone almost as strong. Restraint isn’t kindness. It’s a strategy that wins over time, because the cost of finishing off a rival can exceed the benefit, even for the side that would win.

The same logic explains one of the stranger moments in tech history. In August 1997, with the Justice Department actively investigating Microsoft for antitrust violations tied to Windows’ dominance, Bill Gates announced a $150 million investment in Apple. This was Microsoft’s most visible rival, and a company that was, at the time, weeks from running out of cash. Microsoft didn’t need Apple to survive. But a market with no visible alternative to Windows was a much easier case for regulators to make than a market with a struggling also-ran limping along. Keeping Apple alive was cheap insurance against a far more expensive outcome: a forced breakup.

It’s tempting to read every dominant-firm-tolerates-weak-rival setup this way, and that’s worth resisting. Take Google and Bing. Bing has been struggling under the shadow of Google’s search dominance, and at first glance, this looks like the same pattern. But in August 2024, Judge Amit Mehta ruled that Google is a monopolist anyway, rival or no rival, and testimony in the case suggested there was no price Microsoft could have offered to get Bing preloaded onto Apple’s devices. Bing’s survival looks less like deliberate mercy and more like accidental irrelevance. Even after the ruling, no major distributor has switched to it.

Intel is the sharper cautionary tale, because it shows what happens when a dominant firm chooses the opposite strategy. For most of the 2000s, Intel controlled somewhere between 80 and 98% of the market for PC processors, with AMD as its only real competitor. Rather than tolerate AMD, Intel spent years paying computer makers, through exclusive rebates, not to use AMD’s chips even when they were better and cheaper. That’s suppression, not forbearance. It cost Intel a record $1.45 billion fine in Europe, a $1.25 billion settlement with AMD, and a formal FTC consent decree restricting its conduct going forward. Intel didn’t get punished for having a weak rival. It got punished for trying to finish the job.

There’s a real pattern underneath all three stories, though, and biology has a clean illustration of it. Chimpanzees are territorial, and male chimps do sometimes kill members of a rival group. But this almost always only happens when a raiding party heavily outnumbers its target, which sharply lowers the attackers’ own risk. Restraint, in other words, isn’t a fixed trait. It’s a bet that holds only as long as the cost-benefit math favors restraint. That’s true for a chimp deciding whether to raid. It’s true for Microsoft deciding whether to prop up Apple. And it’s true for a court deciding whether “there’s technically a competitor” still counts as a real defense.

Why this matters for how you analyze a company. This cuts two ways depending on whether you’re evaluating the leader or the spared rival.

If you own the dominant player, the discipline itself is worth watching as a signal. A leader that keeps a weaker rival alive, even at some cost to near-term share, is managing regulatory tail risk cheaply. A leader that pushes for full elimination, or tries to acquire the rival outright, is telling you either that it’s supremely confident the old playbook no longer applies, or that management doesn’t fully understand the game it’s playing. Either way, that’s a decision that tends to show up in a DOJ complaint years before it shows up in a P&L, and it’s worth pricing in earlier than the market usually does. Intel is the clearest evidence of what that costs in practice: a decade of choosing suppression over forbearance ended in billions of dollars in fines and settlements, borne directly by shareholders.

If you own the spared rival, the harder and more useful question is this: how much of this company’s survival is actually its own doing? A standard valuation prices the #2 or #3 player on its own market share, its own unit economics, its own competitive position. But if part of what’s keeping it alive is a decision made by someone else, the real tail risk isn’t sitting in the 10-K. It’s in whatever could change the leader’s calculus: a new CEO with a different risk tolerance, a court that starts scoring specific conduct instead of counting rivals, the leader’s own share drifting to a point where elimination stops looking dangerous. Model the leader’s incentives, not just the rival’s fundamentals, or you’ll misprice the downside case on names like this.

II. Profit Forbearance: Why Stags Roar Instead of Fight

Go back to those red deer for a moment, because the specific mechanism is worth unpacking. Fighting is genuinely costly for stags. In a given rut, up to 6% of them come away with a permanent injury, and fights occasionally end in death. So rather than escalate straight to antlers, stags spend most of the rut roaring. In a landmark 1979 study, zoologists Tim Clutton-Brock and Steve Albon found that the roaring itself does real work. It’s an honest, hard-to-fake signal of a stag’s size and stamina, and rivals use it to size each other up before deciding whether a fight is even worth having. Only when the vocal contest fails to settle things do stags escalate to the parallel walk, and only when that fails do they actually lock antlers. The costly, dangerous behavior only comes out as a last resort.

Coke and Pepsi run the same play. Their roughly 75% combined share of the U.S. carbonated soft drink market looks, on paper, like exactly the kind of concentration that should worry an investor. But that’s only true until you notice which dimension the rivalry actually runs on. Between 1997 and 2004 alone, the two companies launched 22 new brands between them. They’ve fought since 1975 over blind taste tests. They compete hard for exclusive placement in restaurants, universities, and stadiums. What they don’t do is cut prices at each other. One industry estimate found that a full Coke-Pepsi merger would only raise prices by 16 to 17%. That sounds significant, until you realize it implies that current pricing, with two competitors, is already close to what a single monopolist would charge anyway.

You see the same shape in an industry with nothing else in common with soft drinks. Boeing and Airbus have split the market for large commercial aircraft for decades, and the rivalry is genuinely brutal: billions in R&D, entire new aircraft families built to leapfrog each other, government subsidy fights that have gone all the way to the WTO. What they generally don’t do is win a sale by simply underpricing a comparable jet. Industry analyses consistently describe headline pricing between the two as comparatively stable, even while non-price competition (innovation, delivery slots, financing, customer service) stays fierce. Different product, different customers, same underlying math.

Why this matters for how you analyze a company. This is the one that most directly changes how you should read a “competitive intensity” narrative. Investors routinely treat visible rivalry (heavy ad spend, a wave of product launches, aggressive marketing) as evidence that margins are under threat. Sometimes that’s right. But if that activity is happening on a different axis than price, it can mean almost the opposite. Competition has been successfully rerouted away from the one dimension that would actually hurt everyone’s margins. Loud, expensive-looking rivalry can be a sign of stability, not fragility. Think of it as two stags roaring at each other instead of fighting.

The practical version of this: before you discount a name for “intense competition,” check what dimension the intensity is actually on. And treat the transition from cheap signaling to costly escalation as the real event to watch for, not the noise leading up to it. A stag doesn’t lock antlers until the roaring fails to settle things; a company doesn’t start a price war until marketing and product launches stop working. So that transition is the real signal that the old equilibrium has broken. The volume of the roaring itself tells you much less.

III. Territorial Forbearance: Why Neighbors Don’t Raid the Border

Territorial animals (birds especially, but also fish, lizards, even ants) behave very differently toward a stranger than toward the animal living on the other side of a boundary they’ve held for a while. Ethologists call it the “dear enemy effect,” first named by James Fisher in 1954 and studied extensively since. Once two neighbors have settled a shared border, they scale back aggression toward each other specifically, while staying just as hostile toward an unfamiliar intruder. It functions almost like an unspoken agreement, and in several species, breaking it has a real cost. An animal that pushes past an established border into a familiar neighbor’s territory tends to get more intrusions back on its own turf afterward, not fewer. The truce holds because violating it is expensive, not because anyone’s being polite.

This is close to a word-for-word description of how airlines behave in their fortress hubs. American Airlines dominates Dallas-Fort Worth; Delta dominates Atlanta. Neither seriously contests the other’s home turf, not out of courtesy, but because both know an attack in one city invites a counterattack in the other. Economists William Evans and Ioannis Kessides documented exactly this pattern across the U.S. airline industry in a 1994 study. A similar effect shows up in banking, where research has found that banks with more overlapping footprints across local markets tend to be more profitable on average. That’s consistent with the same reduced-rivalry logic operating quietly across dozens of local skirmishes at once.

It shows up somewhere less obvious, too: home internet. A study of U.S. broadband providers, using years of real-world speed-test data, found that download speeds are slower in local markets where the same ISPs compete against each other repeatedly. That’s consistent with those providers quietly under-investing in service quality specifically where they keep meeting the same rival. Multimarket contact doesn’t just soften price competition. It can soften the incentive to invest at all.

Why this matters for how you analyze a company. This gives you a specific, checkable due-diligence flag. When a company with meaningful multimarket overlap announces an expansion into a rival’s strongest market, most analysts model the competitive response in the new market: will the incumbent cut prices, ramp up marketing, and so on. Fewer model the retaliation risk back home. But if the expanding company and the incumbent it’s provoking meet each other in several other markets too, here’s the more important question: what happens to the expanding company’s own fortress markets once this starts? The dear enemy effect predicts the incumbent won’t just fight back locally. It’ll push back wherever the two companies already compete. That’s a real cost that rarely makes it into the market’s initial reaction to an expansion announcement, and it’s worth underwriting explicitly before getting excited about a growth story that involves poking a multimarket rival. And don’t treat this as a pricing story only. The broadband finding says the same restraint can show up in capex and service quality years before it ever shows up in a price sheet, which makes it a genuinely useful thing to check in a company’s most overlapping markets before you assume its infrastructure spend there is competitively disciplined.

The Legal Line

None of this requires anyone to talk to anyone, and that’s not a technicality. It’s the entire reason it’s legal. A stag doesn’t sign a treaty with its rival; it independently calculates that fighting isn’t worth it this time. Coke and Pepsi don’t have a handshake deal to leave price alone; each side just knows, every quarter, what a price war would cost, and neither pulls the trigger. That’s the line between tacit restraint and actual collusion, and it’s a real legal boundary, not a semantic one. The moment forbearance becomes an explicit agreement, it stops being smart strategy and becomes a crime. In 2015, Apple, Google, Intel, and Adobe paid $415 million to settle a lawsuit after it came out that they’d made actual bilateral agreements not to poach each other’s engineers. This wasn’t a shared instinct. It was an agreement, in writing, and that’s what made it illegal. Courts have consistently held that companies independently arriving at similar, competition-softening behavior isn’t illegal on its own. Companies agreeing to do it is.

Putting the Models to Work

None of the three patterns above are about animals, or companies, being nice to each other. They’re all the same cost-benefit math, run in a different currency: injury risk for a stag, antitrust exposure for a tech company, a fare war for an airline. What changes is only the stakes on the table and who’s keeping score.

For a long-term investor, that’s the real payoff. “This market looks like a stable duopoly” is a hunch. It’s hard to test and easy to be wrong about. “This market is stable because of a specific, named mechanism, and here’s the exact thing that would break it” is a thesis with a trigger attached, and you can actually watch for the trigger. The next time you’re looking at a dominant company that hasn’t finished off a weaker rival, don’t just note that the rival survived. Ask what’s actually keeping it alive, and whether that reason is still true today. I’d rather be wrong about a named mechanism I can revisit than right about a vague feeling I can’t.