The Innovator's Dilemma

Christensen studied disk-drive makers across several product generations, then checked the pattern in excavators, steel and retail. His finding: leaders almost always won the races for better technology inside their existing market (sustaining innovation), and almost always lost when a cheaper, simpler product arrived that their best customers did not want yet (disruptive innovation). The newcomer starts at the bottom of the market or in a market nobody served, improves faster than customers' needs rise, and eventually becomes good enough for the mainstream — at a cost structure the incumbent cannot match without cannibalising itself.

Why it matters now: the 2026 debate about AI agents versus seat-priced software is a textbook case to test the framework on. An agent that does a slice of the work for a fraction of the price looks inferior to a full product suite — which is exactly the profile Christensen describes. The framework does not say who wins; it gives a structured way to ask which business models sit in the path.

The working use is an exposure check on business models, not a hunt for the next disruptor. For each holding the questions are concrete: are its best customers already over-served, is there a cheaper product entering from the low end or from non-consumers, how fast is that product improving relative to what customers need, and would a response cannibalise the gross margin the business is valued on? The mistake it prevents is reading a high-margin incumbent's strong recent quarters as proof of safety — in Christensen's pattern, the numbers usually look best just before the crossing. It also guards against the opposite error: calling every new entrant disruptive when it is really a sustaining improvement the incumbent will absorb.