The Next Bottleneck Only Matters If Someone Can Own It

The abundance misconception

There is a comfortable idea in strategy circles that goes like this: when something becomes abundant, something else becomes scarce, and the smart move is to go own the new scarce thing. Cheap intelligence means attention gets valuable. Cheap content means trust gets valuable. And so on.

We went and checked. The study behind this piece looked at 21 abundance shocks going back to the printing press: rail, electrification, containerization, PCs, search, cloud, smartphones, digital photography, streaming, e-commerce, plus a deliberately awkward set of cases where the cost collapse was enormous and nothing much happened afterward. LED lighting, DRAM, NAND flash, solar panels, generic drugs, long-distance telephony, airline deregulation, the fiber glut, the Green Revolution.

The controls were the point. It is easy to assemble a list of famous cases where abundance minted a new monopolist, because those are the cases people write books about. If you only look at those, you will conclude that abundance always creates a new bottleneck. The record is less obliging.

Roughly six in ten of the cases we could resolve showed real value migration. The other four in ten showed cheaper prices, more consumption, and customers keeping the gains. Not a weak effect on the losing side either. Some of the largest technological achievements in the sample produced almost no durable rents for anybody.

So the honest version is: abundance does not automatically create a valuable new bottleneck. Most of the time it creates customer surplus. The cases where value migrates share a recognisable structure, and once you can see the structure you can usually tell which situation you are in.

What separates the two outcomes

Start with the cases where nothing migrated, because they are more uniform than the winners.

LED lighting got about 90% cheaper per lumen every decade. Light today is something like 2,000 times more affordable than it was in 1902. Solar modules went from around $100 per watt in 1980 to under $0.10 today. DRAM prices per gigabyte fell by a factor of ten every five years for decades. These are among the steepest cost curves ever recorded in any industry.

The manufacturers made almost nothing. In 2024 all five of the largest solar manufacturers posted losses. Jinko lost around $470 million, LONGi around $1.3 billion. Polysilicon fell from $32.70 a kilo in early 2023 to $4.40 by mid-2024. DRAM and NAND run the same cycle on a repeating loop.

Four separate industries, one structure. Scarcity kept showing up, but none of it stayed scarce enough. Capital and capacity responded, competitors entered, and the margins disappeared. There was no adjacent layer customers had to pass through, nothing that accumulated, nothing anyone could gate. So the surplus went where surplus goes when nobody can hold it: to the buyer.

Long-distance telephony and airline deregulation got there by a different route and landed in the same place. Real fares fell 45 to 50%, passenger volumes more than doubled, and the industry has cumulative losses to show for it. Deregulation was the abundance shock. Competition did the rest.

Generic drugs look like they belong on this list and only half do, which is worth flagging now because it turns out to be the most instructive case in the study. I will come back to it.

Now the other side.

When PC hardware commoditised in the early 1990s, the box makers went to thin single-digit margins and buyers got cheap computers. The money did not disappear. It moved one layer down and one layer up, into the CPU and the operating system. Intel ran 50 to 60% gross margins while other chipmakers ran 3 to 22%. For most of that decade Intel and Microsoft between them earned more than the entire rest of the world PC industry combined. Cheap hardware made the standard more valuable, not less, and the standard was not something a competitor could add capacity to.

Electrification is the physical version of the same thing, which matters because it means this is not simply a digital phenomenon. Electricity became cheap and universal, and the thing that became more important was the wires. You cannot build a second grid down the same street. Entry is constrained by land, rights of way and regulation, so the position is durable. It has been durable for a century, which is longer than any digital position in the sample. Regulated utilities earn steady returns rather than spectacular ones, which is the trade: the same regulation that protects the position also caps the rent.

Streaming did it with rights. Distribution went to roughly zero, which should have been terrible news for the music industry, and initially was. What actually happened is that as distribution stopped being scarce, the catalog became the only thing left that was. Rights holders now take around 80% of the value of an audio stream. Spotify, the aggregator everyone thinks of as the winner, runs thin. The label that owns a back catalog nobody can recreate does better than the platform that delivers it.

Digital photography is the sharpest version. Marginal cost per photo went to zero. Kodak went from $16 billion in revenue to Chapter 11. Camera shipments fell 94% from their 2010 peak. Enormous surplus to consumers, who now take something like 1.6 trillion photos a year. But value did not vanish, it relocated twice. It went to the attention platform where the photos ended up, and Facebook bought Instagram and its 13 employees for a billion dollars in 2012, the same quarter Kodak filed. And it went to image sensors, where Sony holds something around half the smartphone market. Two adjacent layers, both hard to enter, both more valuable precisely because photos got free.

Strategic importance is not the same as economic value

This is the part I would want an executive to actually internalise, because it is where most of the expensive mistakes live.

Fiber was indispensable and worthless to own. Container shipping cut the cost of moving a ton of cargo by 97%, remade global trade, and container carriers have spent the decades since going bankrupt in cycles. Generic drug manufacturers keep the healthcare system functioning at margins that would embarrass a distributor. Solar manufacturers are decarbonising the planet at a loss.

All of these are strategically critical. None of them have bargaining power.

Meanwhile the value in generic drugs did migrate, just not to anyone making drugs. It went to the pharmacy benefit managers. Three of them, CVS Caremark, Express Scripts and Optum Rx, processed about 80% of American prescription claims in 2024. They do not manufacture anything. They sit between a commoditised product and the patient and decide what gets on the formulary. Commoditisation upstream made the gatekeeper more valuable, not less.

That is the pattern. When something becomes abundant, look one step along the chain for the thing demand is now growing faster than supply can respond to, then ask who controls it.

So the question is not "is this asset becoming more important." Plenty of assets become more important and stay bad businesses. The questions that actually predict the outcome are narrower:

Can customers substitute away from it? Can competitors add supply within a few years? Does one company control access, or is access open? Does the position get stronger with use, or does it just sit there? Can regulation cap or remove the rent? And when things get tight, does the customer pay you more or use you less?

That last one is the cleanest tell. When bandwidth got scarce nobody paid fiber owners more, because there was 20 times more of it than anyone needed. When app distribution got valuable, developers paid Apple 30% and had no alternative.

Nothing digital is permanent either

Two of the strongest positions in the sample eroded, which is worth sitting with before assuming a good position stays good.

Wintel held for about twenty years and then mobile happened. The standard was still the standard, it just stopped being the standard that mattered, and value moved to phones and cloud. Nobody competed with Windows. Something else got built next to it.

The App Store is the other one. Apple's 30% commission was one of the great distribution positions ever assembled, and it held cleanly for more than a decade. Then the Digital Markets Act and the Epic ruling forced alternative payment routes and commissions started coming down. No competitor beat Apple. A regulator changed the rules.

Regulation cuts both ways in this data. It created durable positions in utilities, spectrum and pharmacy benefits. It destroyed them in railroads and app stores. If your position depends on a gate, the gate can be legislated open, and that risk goes up roughly in proportion to how well the position is working.

The version you can actually use on Monday

Take whatever is getting cheap in your market and walk it through five questions in order. It takes about twenty minutes and it is more useful than most strategy offsites.

The strategic error this framework is designed to catch is the common one: investing in the layer with the most visible demand growth rather than the layer that will keep the economics. Everyone who built fiber in 1999 was looking at a demand curve going straight up.

Practically, this pushes you toward a small number of moves. Secure the constrained physical or legal thing early, because that is the one you cannot buy later at any price. Own the record of what happened rather than the thing that does the work. Take exclusivity where exclusivity is available, since non-exclusive access to anything is a service, not a position. And be willing to let go of the layer that is heading for commodity economics, rather than defending it because it used to be the business.

The uncomfortable part

If AI is making something you currently sell dramatically cheaper, the instinct is to defend the old scarcity. Add features, tighten the contract, argue that your version is better.

Kodak had the best film in the world.

The more useful question is what your customers are going to depend on next, whether that thing can be built by anyone with capital and a couple of years, and whether you can own enough of it to matter. If you cannot own it, do not build the strategy on it. Sell into it instead and let someone else fund the buildout.

So: what is becoming abundant around you, and what does that make newly important that you could actually control?

If you can answer the second half of that with something specific, you have a strategy. If you can only answer the first half, you have a trend.