Your Favorite Compounder Might Be a Melting Ice Cube

In 1975, Kodak controlled nearly 90% of the U.S. film market.

By 2012, it filed for bankruptcy.

The strange part? Kodak invented the digital camera in 1975—the very technology that killed it.

This isn't a story about stupidity. It's a story about how competitive advantages decay. And the pattern Kodak followed isn't rare. It's the default.

Most investors understand that moats matter. Fewer understand that moats erode—often invisibly, often years before the earnings show it.

This essay maps the mechanics of moat decay: the shapes it takes, the metrics that reveal it early, and the capital allocation traps that accelerate it.

By the end, you'll have a framework for distinguishing a durable franchise from a melting ice cube dressed up in record earnings.

Here's the uncomfortable truth about high returns on capital: they attract competition the way blood attracts sharks.

A company earning 30% ROIC is holding up a sign that says "steal this business."

Capital flows toward excess returns. Competitors build "good enough" alternatives. Venture subsidies fund customer acquisition wars.

The incumbent responds by spending more on marketing, on R&D, and on retention. Incremental returns decline. Margins compress.

Eventually, ROIC converges to the cost of capital. Economic profit hits zero.

This is mean reversion. It isn't a bug in capitalism. It's the operating system.

Morgan Stanley data shows that companies in the top quintile of ROIC see roughly half their excess returns disappear within five years. The gap closes faster than most investors expect.

Not all moats erode the same way. The shape of the decline determines how much warning you get—and how dangerous the terminal value really is.

Linear Decay

This is the slow fade. Brand-based moats often follow this pattern.

Kraft Heinz is the textbook case. In 2019, the company wrote down $15 billion in brand value. The moat didn't collapse overnight—it thinned over years as private-label alternatives closed the quality gap.

Coca-Cola shows a subtler version. Volume growth has been flat-to-negative for over a decade. The company maintains earnings through price increases and cost cuts. But pricing power without volume growth is a moat on borrowed time.

Step-Function Decay

This is the cliff. One technological shift, one regulatory change, and 50–80% of the value disappears in a single cycle.

Nokia held over 40% of global smartphone market share in 2008. By 2013, it was under 3%. The iPhone didn't slowly erode Nokia's position—it made the entire keyboard paradigm obsolete.

BlackBerry followed the same arc. Its enterprise security moat meant nothing once executives wanted touchscreens. The switching cost was high—until it wasn't.

Intel is the most recent example. For decades, it dominated x86 manufacturing. Then TSMC pulled ahead in process technology, Apple designed its own chips, and the ecosystem shifted to ARM. Intel's ROIC went negative. The moat didn't erode. It step-functioned.

Exponential / Reflexive Decay

This happens when the moat feeds on itself in reverse.

MySpace once had 100 million users. But network effects work both ways. When early adopters migrated to Facebook, the platform became less valuable for everyone who stayed. The exodus accelerated.

Tumblr was sold to Yahoo for $1.1 billion in 2013. Six years later, Verizon sold it for under $3 million. The creative community left. The network hollowed out. Reflexive decay.

Social platforms and liquidity-dependent marketplaces are especially vulnerable here. The same flywheel that built the moat can spin in reverse.

Here's where it gets tricky.

A company can report record EPS while its moat is already breached.

Management has tools: they can increase leverage, cut R&D, defer maintenance capex, or harvest the brand. These moves boost short-term earnings. They also accelerate long-term decay.

General Electric did this for a decade. EPS kept growing. Industrial owner's earnings diverged from GAAP net income. The rot was hidden in the capital structure.

The key insight: ROIC is a lagging indicator.

By the time declining returns show up in the financials, the structural moat—the thing that created those returns—may already be gone.

To see decay early, you need to track what Buffett called "owner's earnings": net income plus depreciation, minus the true maintenance capex required to hold a competitive position.

If that maintenance capex is rising while market share is flat, the moat is thinning. Regardless of what the P&L says.

Switching costs are often the most durable moat. Deeply integrated enterprise software. Medical billing systems. Core banking infrastructure.

But there's a hidden fragility.

Switching costs don't protect forever. They protect until the pain of switching is exceeded by the cost of staying.

When that threshold is crossed, decay is rapid and total.

The shift toward API-first architecture is lowering switching costs across enterprise software. Integration is becoming modular. The lock-in is loosening.

Intel is a case study. For decades, it held a near-monopoly on x86 chips. The switching costs were immense—software ecosystems, manufacturing partnerships, design dependencies.

Then the ecosystem shifted. ARM architectures rose. TSMC pulled ahead in manufacturing. Apple designed its own silicon.

Intel's ROIC went negative. Margins halved relative to competitors. The moat didn't erode slowly. It step-functioned.

Some moats do defy gravity.

Morgan Stanley's data shows that 25–30% of high-ROIC companies sustain their excess returns beyond the typical five-year fade window. Microsoft in the Nadella era is a recent example—revenue doubled, margins expanded, the moat widened.

There may be a "Lindy Effect" at work: the longer a moat has survived competitive assault, the longer it may survive in the future.

But this is also where hindsight bias creeps in. It's easy to identify the survivors after the fact. Harder to predict which moats will compound and which will collapse.

The honest answer: decay is the default, but not the destiny. The question is whether management is building the moat—or harvesting it.

Quantitative Signals:

Qualitative Red Flags:

The Death Cross of Capital Allocation: When incremental ROIC falls below the cost of capital, every dollar management reinvests destroys shareholder value—even if reported earnings are growing.

This is the silent killer. The financials look fine. The economics have already turned.

A moat is not a wall. It's a delay mechanism.

The value of a business is the present value of the cash flows generated during its Competitive Advantage Period. When that period shortens—even by a few years—terminal value can collapse faster than near-term earnings estimates suggest.

The best investors don't just ask, "Does this company have a moat?"

They ask: How fast is the moat evaporating? And does management know?

Most don't. Most keep reinvesting in a dying castle.

The few who recognize the decay early—and pivot the cash flow elsewhere—create enormous value. Berkshire's shift from textiles to insurance. Nokia's pivot back to network infrastructure.

The real edge isn't finding a moat. It's knowing when the moat has already started to melt.