What the Innovator's Dilemma Actually Says
In 1997, a Harvard Business School professor named Clayton Christensen published a book with a deliberately paradoxical title: The Innovator's Dilemma: When New Technologies Cause Great Firms to Fail. It won the Global Business Book Award for the best business book of the year, and it changed how a generation of founders and executives think about competition.
The dilemma is this. The decisions that keep a great company great are the same decisions that leave it exposed. A well-run firm listens to its best customers, invests in higher-margin products, and improves the things those customers ask for. That's textbook good management. It's also the reason market leaders get blindsided by cheaper, simpler competitors they were right to ignore, until suddenly they weren't.
Christensen, who lived from 1952 to 2020, wasn't writing a story about lazy or arrogant managers. That's the part most people get wrong. His companies fail because they're well managed, not in spite of it. The rational choice, made by smart people with good data, is often the fatal one.
That single reframe is why the book still gets cited by founders decades later. It turned "why did that giant collapse?" from a story about incompetence into a story about incentives. And once you see the pattern, you start noticing it everywhere.
The Disk Drives That Started It All
Christensen didn't start with a grand theory. He started with hard drives, and he had a good reason. The disk drive industry moved so fast, with so many product generations packed into so few years, that he compared it to fruit flies, borrowing a line from a friend. If you want to study how species rise and die, you watch fruit flies because they cycle through generations quickly. Disk drives let Christensen watch companies rise and die the same way.
The pattern repeated with almost mechanical regularity. The industry moved from 14-inch drives to 8-inch, then 5.25-inch, then 3.5-inch. Each shrink was a chance to build a smaller, cheaper drive for a new kind of computer: mainframes, then minicomputers, then desktop PCs, then laptops.
Here's the strange part. At nearly every transition, the established leaders lost. They had the engineering talent to build the smaller drives. Many of them did build prototypes. But their biggest, most profitable customers didn't want smaller drives yet, so the projects got starved of resources. New entrants, with nothing to lose and a fringe market to chase, took the new architecture and rode it upmarket.
Christensen's central chapter, "How Can Great Firms Fail? Insights from the Hard Disk Drive Industry," laid out the data. Established firms consistently led on sustaining improvements, the ones their customers asked for. They just as consistently lost the disruptive transitions, the ones their customers didn't yet value. The technology wasn't the problem. The customer relationship was.
Sustaining vs Disruptive Innovation
The heart of the theory is a distinction that sounds simple and turns out to be profound. Not all innovation is the same. Christensen split it into two kinds that behave in opposite ways.
A sustaining innovation makes an existing product better along the dimensions customers already care about. A faster processor, a sharper camera, a car with better mileage. Incumbents almost always win these battles, because they have the customers, the resources, and every reason to keep improving.
A disruptive innovation is different. It shows up worse on the metrics the mainstream cares about, but better on something else, usually price, simplicity, or convenience. It serves customers the incumbent doesn't want or can't reach. And it improves fast enough that, eventually, it's good enough for the mainstream too.
| Dimension | Sustaining Innovation | Disruptive Innovation |
|---|---|---|
| Performance | Better on established metrics | Worse at first, better on a new metric |
| Target customer | Existing, most demanding customers | Overlooked, low-end, or non-consumers |
| Price | Same or premium | Lower, often much lower |
| Who usually wins | The incumbent | The new entrant |
| Incumbent's reaction | Invest aggressively | Ignore, then react too late |
The trap lives in that last row. Incumbents pour resources into sustaining innovations because the math is obvious and the customers are asking. They dismiss disruptive ones because the math looks terrible and the customers are nobody they recognize. Both decisions are rational. Only one of them is survivable.
Why Great Companies Fail by Doing Everything Right
So why can't a smart incumbent just fund the disruptive product? This is where Christensen's answer gets uncomfortable, because it isn't about willpower. It's about how resources actually get allocated inside a healthy company.
Think about what happens when a promising engineer pitches a cheap, low-margin product for a tiny market nobody's proven yet, right next to a proposal to improve the flagship product your biggest customers are demanding. Every incentive, from quarterly targets to the sales team's commissions to the CFO's margin goals, points at the flagship. The disruptive project loses the internal fight, every time, on the merits.
Christensen called the surrounding web of customers, suppliers, and expectations a value network. A company gets so tuned to its network that it literally can't see value where its best customers see none. The organization's strengths, its focus and discipline and customer obsession, become the exact reasons it can't respond.
There's a second problem: size. A billion-dollar company needs big new markets to move the needle. A market that starts at ten million dollars is a rounding error, not worth the attention. But disruptive markets always start small. By the time one is big enough for the incumbent to care, the entrant has years of head start and a product that's no longer worse.
This is the dilemma stated plainly. Do what made you successful, and you miss the disruption. Chase the disruption, and you starve the business that pays the bills today. Neither path is obviously right when you're standing at the fork, which is why so many great companies pick wrong. The same tension shows up in why competition is for losers, where the safest-looking market position is often the most fragile.
Low-End and New-Market Disruption
Christensen and his co-author Michael Raynor refined the theory in a 2003 follow-up, The Innovator's Solution. One of the most useful additions was splitting disruption into two distinct paths, because they don't look the same when they start.
Low-end disruption attacks customers the incumbent has overshot. Products keep improving faster than most people's needs, so eventually the mainstream product is more than good enough, and overpriced for what a big chunk of buyers actually want. A cheaper, simpler rival swoops in at the bottom, takes the least demanding customers, and then climbs.
New-market disruption is sneakier. It doesn't steal existing customers at all. It creates a market among people who were priced out or shut out entirely, the non-consumers. Because it competes against "nothing," incumbents don't even register it as competition until it's grown large enough to pull mainstream customers into the new way of doing things.
| Feature | Low-End Disruption | New-Market Disruption |
|---|---|---|
| Who it targets | Overserved existing customers | Non-consumers who had no good option |
| The pitch | "Same job, much cheaper" | "Now you can do this at all" |
| Incumbent blind spot | Happy to shed low-margin customers | Doesn't see it as competition |
| Classic example | Steel minimills, discount retail | Personal computers, digital cameras |
| End state | Climbs upmarket into the mainstream | Pulls the mainstream into a new market |
Most real disruptions blend the two. The key insight is the same either way: disruption almost never arrives as a head-on assault. It comes from below or from the side, from a direction the incumbent has good reasons to ignore.
Case Studies: Kodak, Blockbuster, and the Steel Mills
Theory is cheap. What makes the innovator's dilemma stick is that the same script has played out, with real companies and real dates, again and again.
Kodak invented its own killer. In 1975, a young Kodak engineer named Steven Sasson built the world's first digital camera, a toaster-sized device that captured a 0.01-megapixel black-and-white image onto cassette tape. He showed it to management. As Sasson later recalled, the reaction was, "That's cute, but don't tell anyone about it." Kodak made its money on film, and digital photography threatened the most profitable business the company had. So it shelved the prototype and kept protecting film long after the shift was obvious. In January 2012, Kodak filed for Chapter 11 bankruptcy, undone by the exact technology its own engineer had prototyped 37 years earlier.
Blockbuster passed on Netflix. Netflix launched in 1997 as a DVD-by-mail service, a clunky, mail-order alternative to driving to a store. In 2000, Netflix co-founders Reed Hastings and Marc Randolph reportedly offered to sell the company to Blockbuster for around 50 million dollars. Blockbuster, then the giant of video rental with thousands of stores and healthy late fees, wasn't interested. By 2010, Blockbuster filed for bankruptcy, roughly a billion dollars in debt. Netflix crossed 20 million subscribers the same year and never looked back.
Nucor climbed the ladder in steel. The cleanest low-end disruption in the book is the steel minimill. Companies like Nucor started at the very bottom, melting scrap into cheap, low-quality rebar that the big integrated mills were glad to abandon because the margins were awful. By 1980, minimills had captured about 90 percent of the rebar market. Then they improved, moved up into bars and rods and angle irons, and kept climbing toward the high-end sheet steel the incumbents thought was safe. Each rung looked like a small loss to the incumbents. Together they were the whole ladder.
The Nucor story has a sequel that's just as instructive. When Christensen shared the minimill research with Intel's Andy Grove, Grove got the point immediately. "Rebar" became Intel's internal slogan for attacking the cheap PC market from below before a rival could. Intel launched the low-cost Celeron chip in 1998 to defend its own low end. That's the innovator's dilemma solved, not by ignoring the threat, but by deliberately building the cheap, "worse" product before someone else did.
When Christensen Got It Wrong
A theory that explained everything would explain nothing, and Christensen's has real limits. Being honest about them makes the framework more useful, not less.
The most famous miss is the iPhone. In 2007, Christensen predicted it would fail. His reasoning followed the theory: the iPhone was a sustaining innovation relative to Nokia's dominance, an improvement on existing phones rather than a cheap disruptor from below, so incumbents should win. "The prediction of the theory would be that Apple won't succeed with the iPhone," he told BusinessWeek. "History speaks pretty loudly on that." The iPhone went on to generate roughly 150 billion dollars in revenue in its first five years. The theory looked at the phone market and missed that the iPhone was really disrupting the laptop, a new-market disruption in disguise.
In 2014, historian Jill Lepore published "The Disruption Machine" in The New Yorker, a pointed critique of the whole framework. She argued the theory relied on circular reasoning and cherry-picked cases. If a firm gets disrupted it proves the theory, and if it survives that proves it too. She pointed out that Seagate, one of Christensen's supposedly doomed disk drive makers, was still alive and shipping billions of drives. The debate that followed sharpened an important point: disruption theory is a lens, not a law of physics.
Christensen spent his later years fighting a different problem, the word itself. "Disruptive" had come to mean "new and threatening," which is not what he meant at all. In a 2015 Harvard Business Review article, "What Is Disruptive Innovation?", he and his co-authors argued that Uber, the poster child for disruption, doesn't actually fit. Uber started in the mainstream market and appealed to existing taxi customers, rather than climbing up from the low end or serving non-consumers first. Calling everything disruptive, he warned, turns a precise idea into a buzzword.
Disruption in the Age of AI
The innovator's dilemma reads like it was written for the current moment. Swap disk drives for AI models and the plot barely changes.
Look at the low-end pattern. A wave of cheaper, smaller, "good enough" AI models keeps arriving. They're worse than the frontier systems on the hardest benchmarks, which is exactly why the incumbents serving demanding enterprise customers can dismiss them. But most tasks don't need the frontier. Once a cheap model is good enough for the everyday job, the expensive one starts to look overshot, the way a mainframe-grade disk drive looked overshot to a laptop maker.
Now look at the new-market pattern. AI tools are pulling in people who never touched the old software at all: non-coders shipping apps, solo founders running functions that used to need a team, students learning in ways no classroom offered. They aren't stealing the incumbent's customers. They're creating new ones, which is precisely why the threat is easy to underrate until it's everywhere. We explored where this leads for software in the SaaS apocalypse and AI-native software, and what it means for tiny teams in the three-person unicorn.
The incumbents face Kodak's choice in real time. Cannibalize your profitable product with a cheaper AI-native version, or protect the margins and hope the disruptor stays small. The companies studying the dilemma are trying to be Intel with its "rebar" strategy: build the cheap, self-disrupting product on purpose, before a startup builds it for them. Whether they can overcome their own value networks is the trillion-dollar question of this decade.
How to Spot Disruption Before It Hits You
Here's the part that matters for anyone who isn't running a Fortune 500 company. The innovator's dilemma isn't only a lesson for incumbents. It's a lesson about attention. Disruption is always visible before it's obvious. Someone sees it early. The question is whether that someone is you.
The people who catch disruptive shifts first aren't smarter. They're paying attention to the right edges: the cheap tool power users are hacking together, the "toy" product a fringe group loves, the workflow that shouldn't work but does. Incumbents miss these signals because their attention is locked on their best customers. You don't have that constraint, but you do have a different one: the signal is buried in an overwhelming stream of articles, videos, and posts.
That's a knowledge management problem, and it's solvable. A deliberate reading habit turns scattered noise into an early-warning system:
- Highlight the anomalies, not just the headlines. When you read about a "worse but cheaper" product or a tool serving people the market ignores, mark it. Use Glasp's web highlighter to capture the passage in context instead of losing it to a closed tab. Over months, your highlights form a map of where disruption is brewing.
- Mine the founders and analysts on video. A huge share of real-time strategic thinking lives in podcasts and talks, not articles. Run them through YouTube Summary to pull the key claims and timestamps, so an hour of founder conversation becomes a few minutes of highlights you can actually keep.
- Interrogate your own archive. Patterns hide across sources. Ask Glasp's AI chat questions across everything you've saved, like "what cheaper alternatives have I flagged in this industry?" That's how weak signals from ten different articles snap into a single trend.
- Watch what others are watching. Disruption is a collective signal before it's a personal one. Glasp's community feed shows what curious people are highlighting right now, which surfaces the fringe topics gaining momentum before they hit the mainstream.
This is the same discipline that shows up in how the best founders build knowledge systems, and it pairs naturally with Clayton Christensen's Jobs to Be Done theory: disruption tells you where change comes from, and Jobs to Be Done tells you why customers switch when it arrives. Track both, and you stop being surprised.
Frequently Asked Questions
What is the innovator's dilemma in simple terms?
It's the trap where a successful company fails precisely because it makes smart, customer-focused decisions. By listening to its best customers and investing in its most profitable products, a market leader ignores cheaper, simpler innovations that seem unattractive at first but eventually take over the market. Doing everything "right" is what leaves it exposed.
What's the difference between sustaining and disruptive innovation?
A sustaining innovation improves an existing product on the metrics customers already value, like a faster chip or a sharper camera, and incumbents usually win these. A disruptive innovation is cheaper and worse on the mainstream metrics but better on price or convenience, serves overlooked customers first, then improves until it's good enough for everyone. Entrants usually win those.
Is Uber a disruptive innovation?
By Clayton Christensen's strict definition, no. In a 2015 Harvard Business Review article, he argued Uber doesn't fit the model because it started in the mainstream taxi market and appealed to existing customers, rather than starting among non-consumers or at the low end and climbing up. It's a successful business, but "disruptive" has a specific meaning that Uber doesn't match.
Why did Christensen think the iPhone would fail?
In 2007 he saw the iPhone as a sustaining innovation relative to Nokia, an improvement on existing phones rather than a cheap product attacking from below, so his theory predicted the incumbent would win. He was wrong, largely because the iPhone was actually a new-market disruption of the laptop and the personal computer, not just a better phone.
How does the innovator's dilemma apply to AI?
Cheap, "good enough" AI models attacking from the low end and AI tools serving people who never used the old software both follow the classic disruption pattern. Incumbents can dismiss them as inferior right up until they're good enough for the mainstream, which is the same trap that caught Kodak, Blockbuster, and the integrated steel mills.
Conclusion: Read the Signals Early
The innovator's dilemma endures because it refuses to let anyone off the hook. It doesn't blame bad managers or dumb decisions. It shows how brilliant people, following the best available data, walk straight into the trap. That's what makes it frightening, and useful.
You can't out-manage the dilemma with discipline, because discipline is part of the problem. What you can do is widen your attention. Watch the low end. Watch the non-consumers. Watch the "worse" product that a fringe group can't stop using. Those are the early pages of every disruption story ever told.
For a reader, that means building the habit of capturing what you notice before it's obvious to everyone. Start highlighting the anomalies with Glasp's web highlighter, turn the founders and analysts you follow into searchable notes with YouTube Summary, and let Glasp's AI chat connect the dots across everything you've saved. Christensen gave us the map. Whether you see the disruption coming is a matter of where you choose to look.