When the moat becomes the anchor

The advantage that goes negative

Mercedes won eight straight constructors’ titles in the hybrid era, and that record is now the reason to bet against them in 2026. The sport is resetting its engine formula to a roughly 50/50 split between combustion and electric power. Every hour Mercedes spent perfecting the old power unit was an hour spent learning a physics that no longer governs the race. Their advantage under the new rules isn’t zero. It runs negative.

That word matters. We’re trained to think of an incumbent as either ahead or, in the worst case, back at the starting line with everyone else. The reset is worse than that. Seven years of hybrid-era telemetry, seven years of engineers who think in the old constraints, seven years of a factory organized around a solved problem: none of it transfers to a 50/50 architecture. Some of it actively misleads. The instinct that was right for a decade becomes the instinct you have to override, and overriding a decade of instinct is slower than not having it.

This is the shape of every substrate change. Clayton Christensen named the incumbent’s disadvantage decades ago, and the point holds beyond disk drives and steel. The deeper you specialized to win the last game, the more of yourself you have to destroy to play the next one.

Specialization is a bet that the rules stay put

Every moat is a wager that the rules stay put. You pour resources into a specific configuration of the world because that configuration pays. The moat is real for exactly as long as the configuration is.

Yahoo’s directory was a genuine asset when the web was small enough to catalog by hand. Editors organized the internet into a tree, and for a few years that was the best way to find anything. Then the web got too big to catalog, search went algorithmic, and the directory became a museum of a smaller internet. The thing Yahoo was best at was the thing that no longer mattered.

Blockbuster’s real estate was the same wager. Thousands of stores near where people lived was the optimal distribution system for physical rentals. Every lease was a vote that video would keep moving through buildings. When distribution went digital, the store count stopped being a network and started being a liability with a monthly payment attached.

Kodak had the deepest chemistry expertise in imaging and a patent wall around film. That was a fortress while images were captured on silver halide. When capture moved to silicon, the chemistry was a sunk cost the company kept paying to defend. Kodak even built early digital sensors and shelved them, because the org could not bring itself to obsolete the asset it was organized around.

None of these companies lacked resources or talent. They lacked the ability to treat their own strength as expendable. Specialization felt like security right up to the moment the substrate moved, and then the same specialization was the anchor.

The challenger starts clean by construction

The challenger’s freedom in a reset isn’t luck. It’s structural. A new entrant has no old physics to unlearn, no factory tuned to the wrong problem, no engineers whose intuition points the wrong way. They read the new rules and build for them directly, which is the only thing the new rules reward.

Audi enters F1 in 2026 with no hybrid-era baggage. Every choice they make is a choice made against the current formula, not the last one. That is not a small head start against a team carrying seven years of the wrong optimization. It is a different starting position on a different track.

Google in 1998 didn’t have to defend a directory. It could treat the entire web as an algorithmic problem because it had never sold anything else. Netflix in 2007 didn’t have to protect store leases when it pushed into streaming, because it had no stores. In both cases the challenger’s advantage was the absence of a thing to protect. Nothing to unlearn, nothing to defend, nothing pulling the org back toward the era that was ending.

The incumbent looks at the challenger and sees someone with less. Less data, less history, less installed base. What the incumbent is actually looking at is someone who is allowed to build for the world that’s arriving instead of the world that’s leaving.

Run two orgs, and don’t let the old one vote

The incumbent play that actually survives a reset is to stop treating the current platform and the next one as a single roadmap. Run two orgs. One keeps the current platform live and supported, funded to defend the position that still pays today. The other is built to win the new rules, and is explicitly told to behave as if the current advantage doesn’t exist.

The second org can’t report to the first. If the streaming team answers to the DVD business, the DVD business will starve it every quarter the numbers get tight, because protecting today’s revenue is a rational local decision that adds up to losing. The whole point of separation is to prevent the current moat from voting on the next one. The org that owns the old advantage cannot be the org that decides how fast to abandon it.

This is expensive and it feels like heresy. You are funding a team whose job is to make your best asset irrelevant, and you are shielding that team from the people who built the asset. Most incumbents can’t do it, not because they don’t understand the logic but because the current platform is where the revenue, the headcount, and the political weight all live. The reset punishes exactly the org that lets its most successful people set the agenda. The problem sharpens when the substrate underneath you is one you rent rather than own rather than rent the layer. The platform can change the rules on you the way a regulator changes an engine formula, and you find out how much of your advantage was borrowed.

The counterargument that fails on unlearning

The strongest objection: the incumbent has the cash and the talent to rebuild for the new rules faster than any challenger. Mercedes can hire, spend, and iterate at a scale Audi’s new program can’t touch. In raw resources this is true, and it is not a small thing.

It fails on the one thing money buys slowly. The bottleneck in a substrate change isn’t building the new capability. It’s unlearning the old one. A thousand engineers who think in the previous formula are slower to retrain than a hundred hired straight into the new one, because you are not filling empty heads. You are overwriting full ones. Cash accelerates building and does almost nothing for unlearning. That’s the asymmetry Christensen kept finding: the incumbent’s resources are real, and they get spent defending the thing that has to die.

Mercedes will not be slow in 2026 because they lack money or brains. They’ll be slow, if they’re slow, because the fastest team on the old rules built an entire organization around being fastest on the old rules. The moat held for eight seasons. The substrate moved. Now the same depth that made them dominant is the weight they have to drag to the new starting line, and the team with nothing to drag gets there first.