“Disruption” has been used to dignify visible violets added to soap, dental floss in three new colours, and rebranded shampoos for so long that the word means almost nothing. The original concept — Clayton Christensen’s specific argument about how new entrants serve previously unmet markets and eventually displace incumbents — is buried under a decade of marketing fiction. This piece reclaims the word, examines what genuine disruption looks like in an age when most business thinking has converged, and identifies the structural conditions under which real disruption still thrives. Now how plays AI in here, as anywhere else?
Explanatorium · The shape of AI thinking
The previous piece in this Think-Smarter thread argued that AI has produced the most striking convergence of business thinking in modern memory. A tendency towards same sources, same (AI) tools, same use cases, same vocabulary, and hence same conclusions. It also argued that originality is the discipline that resists convergence, and that AI used carefully can support rather than flatten genuine thinking. But the question that argument leaves open is the more uncomfortable one. If convergence is the default and originality is the discipline that resists it, what about the rare cases where new thinking does not just resist convergence but actually displaces what came before? What about disruption?
Before answering, the word itself needs reclaiming.
What disruption actually meant — before the corruption
Clayton Christensen’s The Innovator’s Dilemma, published in 1997, made a specific and rigorous argument. Disruptive innovation was not synonymous with novelty, not the same as innovation in general, and certainly not interchangeable with “interesting product.” It described a precise sequence: a new entrant offers a simpler, cheaper, often technically inferior product to a market segment that incumbents have ignored or underserved; the new entrant’s product is initially dismissed by incumbents because it does not meet the needs of their best customers; over time, the new entrant improves their product along the dimensions that matter to mainstream customers; eventually, the new entrant’s product becomes good enough for mainstream use, and the incumbent’s market is captured from below.
That is what disruption meant. Not better marketing. Not faster execution. Not novel features. Specifically the displacement of incumbents by entrants who initially looked too small, too crude, or too irrelevant to take seriously. Christensen’s case studies — the steel mini-mill displacing integrated steel producers, the personal computer displacing minicomputers, the disk drive industry’s repeated cycles of displacement — were structural patterns, not marketing stories. The book’s enduring contribution was showing that incumbents lose to seemingly unimpressive entrants because the incumbents’ rational customer focus prevents them from engaging with markets that look unprofitable from the top.
How the word got corrupted
By 2010 the word “disruption” had escaped Christensen’s specific definition and entered general business vocabulary as a synonym for “noteworthy” or “different” or, most commonly, “marketing-worthy.” Companies described their products as disruptive when they were incremental improvements. Entrepreneurs described themselves as disruptive when they were doing slightly better versions of what already existed. Investors funded startups they called disruptive when they were optimised execution of obvious ideas. The Silicon Valley vocabulary turned “disruption” into a status claim rather than an analytical category.
By 2020, the corruption was complete. Marketing copy described visible violets added to soap as “disrupting the personal care category.” Dental floss in three new colours was “disrupting oral hygiene.” Banking apps with slightly better user interfaces were “disrupting financial services” even when they offered the same products at the same prices to the same customers. The word survived; the concept it once described did not. Christensen himself, before his death in 2020, repeatedly objected to the misuse — pointing out that most things called disruptive in the 2010s were not disruptive by any rigorous definition, and that the conflation made it harder to recognise actual disruption when it occurred.
The corruption is not a minor lexical complaint. It has consequences. When everything is called disruptive, nothing is. When marketing repackaging is treated as equivalent to genuine market displacement, organisations stop developing the analytical capacity to tell them apart. When entrepreneurs claim disruption for products that are merely better-executed versions of existing ones, they crowd out attention from the rare cases that genuinely qualify. The corruption of language causes a corruption of thinking, which causes resources to flow to the wrong places.
Three tests for genuine disruption
If the word is to do useful work again, it needs a stricter definition. Three tests, any one of which qualifies an innovation as genuinely disruptive in the Christensen sense; failing all three relegates the innovation to incrementalism, which is not a criticism but is not disruption either.
Test 1: Does it create a fundamentally different cost structure? Genuine disruption usually involves a new entrant who can serve the same need at a structurally lower cost — not 20 percent cheaper, but an order of magnitude cheaper, because they have eliminated something the incumbent considered essential. The personal computer disrupted minicomputers because it was 100 times cheaper. Cloud computing disrupted on-premises infrastructure because it eliminated capital expenditure entirely. The cost structure must be different in kind, not just better in degree.
Test 2: Does it serve a previously unserved market? Disruptive innovations usually start in markets the incumbents have ignored — too small, too poor, too niche, too unprofitable to bother with. The mini-mill served small construction projects the integrated steel mills found uneconomic. Mobile phones served populations who had never had landlines. The new entrant builds capacity in this overlooked market, then expands toward the mainstream as their product improves. If an innovation is launched for the same customers the incumbents already serve well, it is probably not disruptive — even if it is genuinely better.
Test 3: Does it reframe what the category is for? The deepest disruptions change the question customers ask. Smartphones did not just give people better mobile phones; they reframed “phone” as “computer that also makes calls.” Streaming did not just deliver music differently; it reframed “buying music” as “subscribing to access.” When the question itself changes, incumbents who answered the old question well are structurally disadvantaged. They are competing on metrics that no longer matter.
Innovations that pass none of these tests may still be excellent. They may be highly profitable, well-executed, valuable. They are simply not disruptive in any meaningful sense. Calling them disruptive does not elevate them. It only further empties the word.
So is anything genuinely disruptive in 2026?
Yes, but less than the noise suggests. The honest answer requires distinguishing between the AI tools themselves (which are genuinely disruptive in the Christensen sense — they pass all three tests against the prior generation of cognitive software), and most things being done with AI by businesses (which are convergent applications of those tools, not disruptions of anything).
AI itself disrupts the cognitive software category. The cost structure is fundamentally different — capable language models reduce the marginal cost of writing, summarising, analysing, and reasoning to near zero compared to the human-labour cost they replace. The previously unserved market is enormous: small businesses and individuals who could never afford lawyers, copywriters, analysts, or research assistants now have access to capabilities that were previously restricted to organisations large enough to employ specialists. The category reframing is real: “software that helps you do work” has been replaced by “software that does the work” in a way that breaks the category assumptions of the previous decade.
That is genuine disruption, in the Christensen sense, and it is happening now. But it is happening to the cognitive software category — to the providers of writing tools, research databases, business intelligence software, and so on. It is not happening to most of the businesses that are deploying AI into their existing operations. A consultancy that uses AI to write its reports faster is not disrupting consulting; it is operating more efficiently within consulting’s existing structure. A retailer that uses AI for customer service is not disrupting retail; it is automating customer service within the retail structure that already exists. The disruption is in the AI category itself; what most businesses are doing with AI is convergent adoption, not disruption.
This distinction matters because it changes what the strategic conversation should be about. For an organisation in any industry other than cognitive software itself, the question is not “how can we use AI to be disruptive?” — that question is incoherent in most contexts. The honest question is “how can we use AI to operate well within an industry whose convergence is accelerating?” That is a different question, and it has different answers, and answering the wrong question wastes years of strategic attention.
What does it take for genuine disruption to thrive in late 2020s
Genuine disruption — the kind that passes the three tests — has structural conditions that have not changed because of AI. Five conditions, observable across the disruption cases that have actually qualified over the past five decades.
An incumbent’s blind spot. Disruption requires a market or use case that incumbents are systematically dismissing as too small, unprofitable, or unimportant. The blind spot has to be structural, not accidental — the result of how the incumbent measures success, allocates resources, and rewards its leaders. Without an incumbent blind spot, even excellent new entrants get crushed by the incumbent’s faster response. With one, the new entrant can build capacity unmolested.
A genuinely new cost structure or capability. Christensen’s mini-mills could not have disrupted integrated steel by being marginally cheaper; they were dramatically cheaper. The disruption space requires that the new entrant be able to do something the incumbent fundamentally cannot, not just something the incumbent could match if they tried. AI is opening this kind of gap in many cognitive-work categories, which is why the cognitive software incumbents are visibly struggling.
Time to develop unmolested. Disruption is slow. The new entrant typically needs five to ten years of underestimated, unimportant-looking development before their product is ready to challenge incumbents in the mainstream market. AI may be compressing this timeline somewhat, but the structural pattern still requires a window where the disruptor is small enough to be ignored. Visibility kills disruption; obscurity protects it.
An operator who understands what they are doing. Most disruptions die because the founder thinks they are competing in the incumbent’s market on the incumbent’s terms, and tries to scale up faster than the structural conditions allow. The operators who succeed are usually the ones who understand they are not competing with the incumbent yet, and who patiently build the capacity that will eventually make competition possible. The temperament required is unusual — neither pure ambition nor pure patience, but the disciplined combination of both.
The willingness to be misunderstood. Disruptors are routinely dismissed as toys, niche, irrelevant, or inferior during the years when they are most vulnerable. The operators who eventually displace incumbents are the ones who can hold their nerve through that misunderstanding without abandoning the strategic position that makes disruption possible. The visible violets in soap brigade always sounds more confident in their first three years; the genuine disruptors look uncertain and small while they are building.
These conditions have not changed because of AI. AI may be accelerating the timelines slightly and broadening the categories where disruption is possible, but the structural pattern remains: incumbent blind spot plus new cost structure plus time plus operator discipline plus tolerance for being misunderstood. Operations that meet all five conditions are rare, and they are the only ones that produce genuine disruption. Operations that meet some but not all of the conditions produce successful businesses that are not disruption — which is fine, but should not be confused with the rarer thing.
Avoid
Calling your work disruptive when it isn’t. The honest test takes thirty seconds. Pass none of the three tests, and the word does not apply. Use a different word — “improved,” “differentiated,” “well-executed,” “valuable” — all of which are honest. “Disruptive” used dishonestly does not flatter the work; it only further empties the word for the cases that genuinely qualify.
Mistaking faster execution for disruption. Doing the same thing as your competitors, faster and at lower cost because of AI, is not disruption. It is convergent adoption. It is also good business, in many cases. But it does not qualify as the rarer category, and pretending otherwise leads to the wrong strategic investments.
Believing AI makes everyone potentially disruptive. AI lowers the cost of execution. It does not lower the cost of identifying incumbent blind spots, building genuinely new cost structures, or having the operator discipline to develop capacity unmolested. The structural conditions for disruption have not democratised. The execution capabilities have. Many people who think they are disrupting are merely executing convergent strategies more efficiently — which is fine, but should not be confused with the rarer thing.
Treating soap with visible violets as disruption. Or any of its equivalents. Repackaging is not disruption. Cosmetic differentiation is not disruption. Marketing variation is not disruption. Calling these things disruption is the most common form of the corruption that empties the word. Resist the temptation. The work itself may be excellent. It is simply not the rarer category.
The closing argument
Convergence is the cultural condition of late 2020s business. Originality is the discipline that resists it. Disruption is the rarer outcome that emerges when originality meets the right structural opportunity — the incumbent blind spot, the new cost structure, the time to develop, the operator discipline, the willingness to be misunderstood. The three concepts are linked, and getting any one of them wrong distorts the other two.
The honest version of the AI conversation accepts all three. AI is producing convergence at scale. AI also enables originality for those who use it deliberately rather than reactively. AI is itself genuinely disrupting the cognitive software category, and that disruption is opening secondary opportunities for downstream operators who understand what is actually happening — which is rare. Most businesses applying AI are operating within convergence, which is fine; some are pursuing originality, which is admirable; very few are doing genuinely disruptive work, and the ones that are usually look small and unimportant right now. They will not look that way in five years.
The reductive narrative that AI either enables universal disruption or destroys all originality is wrong in both directions. The honest narrative is more textured and more useful. Convergence is the default. Originality is the discipline. Disruption is the rare outcome. All three are happening. Knowing which you are doing is the strategic question that matters most.
Factbox: the disruption framework, condensed
Origin: Clayton Christensen, The Innovator’s Dilemma (1997). The book identified the structural pattern by which new entrants displace incumbents from below — by serving overlooked markets with simpler, cheaper products that gradually improve toward mainstream use.
The three tests: Genuine disruption requires (1) a fundamentally different cost structure, not marginal improvement; (2) a previously unserved or underserved market, not direct competition with incumbents on their core; (3) a reframing of what the category is for, not just better execution within the category.
The five structural conditions: Incumbent blind spot · genuinely new cost structure or capability · time to develop unmolested · operator discipline · willingness to be misunderstood. All five required; missing one tends to produce successful businesses that are not disruption.
What 2026 actually shows: AI itself is genuinely disrupting the cognitive software category. Most businesses applying AI are in convergent adoption, not disruption. A small number are doing genuinely disruptive work in their own categories using AI as one tool among many. The question of which side of that line you are on is more important than the question of how fast you can adopt.
This piece is the second in a Think-Smarter thread on AI’s effect on business thinking. The first piece — AI convergence: what happens when everyone reaches the same conclusion — establishes the cultural condition this piece extends. The earlier pieces in the broader thread on what is actually happening with AI: The chopper view and The missing piece. For executives navigating these conditions in their own operations, the Gadvisory advisory practice offers structured engagements anchored in this thinking.
© 2026 Think-Smarter.net / Leisure Media Group Ltd. All rights reserved.
