Every few years a generation of well-run companies finds itself outflanked by challengers it chose not to take seriously. It happened to major retailers when Amazon arrived. To Blockbuster when Netflix launched a DVD postal service. To the established banks when Revolut started with a travel card. In each case the incumbent could see what was happening — and rationally decided it was not worth responding to. The challenger was serving a segment too small and too low-margin to justify attention. Every signal pointed toward the current business — and those signals were correct.
Clayton Christensen, the Harvard Business School professor who spent thirty years studying why successful companies lose ground to new entrants, argued that this is not a failure of management. It is a consequence of good management. The standards organisations use to decide what is worth pursuing — what margin justifies investment, which customers matter, what constitutes a real threat — are well suited to running the existing business and poorly suited to seeing what is forming outside it. He called this Values. [1] As organisations grow, values shape thousands of daily decisions without anyone needing to articulate them — consistently directing attention inward, making the opportunity at the edges easy to dismiss until a challenger has already built a position worth contesting.
The challenger enters serving customers the incumbent is not competing for, at margins the incumbent would not consider worth defending, and quietly builds a different kind of competitive position — cost structure, customer relationships in markets the incumbent cannot reach, and an operating model designed for a different logic. As it improves and its trajectory intersects with the mainstream, the incumbent finds that responding effectively would mean working against its own cost structure and margin expectations. Its decisions remain rational until the challenger’s position has grown to the point where an effective response would require dismantling the model that made it successful. [1]
The evidence: it is happening now, in markets considered well defended
Revolut launched in 2015 with one product — a prepaid travel card with better foreign exchange rates than the high street banks. For that specific job it was considerably better than anything the banks offered. But that job was not what the banks’ most valuable customers were asking them to solve. Thin margins, a Lithuanian e-money licence and a segment the banks were not competing for did not register as a threat by any measure they were applying. Their moat — trust, regulatory standing, branch presence, balance sheet — appeared solid.
Over the following decade Revolut added current accounts, savings, trading, crypto, insurance and business banking. It now has sixty-five million customers across a hundred countries, four billion dollars in 2024 revenue — up 72 percent year on year — and a private market valuation of seventy-five billion dollars. Its CEO is targeting nine billion dollars in revenue and three and a half billion in profit for 2026, with a Nasdaq listing aimed at a valuation of at least a hundred and fifty billion dollars. Early investors have seen a twenty-four times return. [2]
Monzo, which followed the same logic, is now the seventh largest bank in the UK by customer numbers. [3] Nubank did the same in Latin America, entering by serving underbanked populations the major banks had no interest in reaching. Nubank now has a hundred and fourteen million customers and reported eleven and a half billion dollars in revenue in 2024, making it the largest neobank in the world by revenue. [4]
Wise built its position in cross-border payments by serving the individual and small business being quietly overcharged on international transfers — a segment too low-value for the banks to prioritise. Western Union, the category incumbent, saw its consumer money transfer revenue decline six percent in the third quarter of 2025. In the same period Wise grew ten percent and Remitly grew twenty-five percent. [4]
Stripe entered payments by serving developers and early-stage startups — a segment the card networks and PayPal were not building for. It processed one and a half trillion dollars in global payments in 2024, up forty percent year on year. Sixty-two percent of Fortune 500 companies now use the platform. [4]
Across all four cases, the incumbent invested in its existing business throughout the period the challenger was building. The issue was never adoption of technology. It was that investment went toward improving what already existed rather than competing in the markets the challengers were entering.
The scale of what has already shifted is instructive. Fintech revenues grew twenty-one percent in 2024 — three times the six percent growth rate of incumbent financial services players. And yet fintechs still only account for three percent of global banking and insurance revenue pools. [4] The disruption is proven and still in its early stages.
Why AI accelerates digital disruption
The digital-native wave disrupted industries protected by physical infrastructure — retail, media, travel, banking. It made limited inroads into the expertise and advisory core of professional services — complex legal advice, professional financial planning, specialist business support, higher education. Delivering that kind of expertise at scale still required expensive human judgment at the centre. The customers priced out of these markets were numerous, but the economics of reaching them did not work without that human expert.
AI is now changing those economics in specific, identifiable markets — not universally, and not without real limitations around accuracy and liability in some contexts. Anthropic’s recent research on AI’s labour market impact illustrates where those markets are. [5] The research compares what AI can theoretically handle across occupational categories against what is actually being used in practice.
Two things stand out. First, AI capability is concentrated in knowledge-intensive occupational categories — legal, healthcare support, financial operations, education, and business and technical roles. These are the areas where the expertise and advisory core of professional services sits. Farming, construction, food preparation and transportation show very limited AI capability. The disruption risk is not spread evenly across the economy.
Second, across every knowledge-intensive category, actual usage is well below what AI is theoretically capable of. The gap between the two is significant — it shows where AI can in principle do the work but has not yet been deployed at scale. That gap maps directly onto the markets Christensen would identify as the entry point: large populations of potential customers currently excluded by price, complexity or availability, in segments incumbents have rational reasons not to build for.
Three broad areas are most immediately exposed. The first is the client-facing advisory layer of professional services. Legal services for individuals and small businesses — contracts, document review, compliance, basic commercial advice — is almost entirely information-based, and the population priced out of affordable legal support is large. Financial planning for the mass market sits below the threshold where professional advice becomes economical for traditional providers; AI is moving into planning, tax and business finance for SMEs in ways robo-advisors never reached. In lending, AI is making viable credit assessment economical for segments traditional lenders find too costly to serve — thin-file borrowers, early-stage SMEs, underbanked populations. In insurance, the broking and advisory layer for the mass market and SMEs is information-intensive and significantly underserved. In healthcare, physical diagnosis and treatment remain largely protected by the requirement for physical presence and regulatory frameworks, but the information and advisory layer — patient triage, chronic condition management, mental health support — is where large populations currently have limited access and where challengers are already active.
The second is marketing, sales and operational support for the mid-market. Marketing strategy and campaign management has long been the preserve of enterprise clients with agency relationships. SMEs and mid-market businesses have been largely priced out. AI-native challengers can now deliver strategy and execution at a price point the excluded market can access. HR and recruitment for smaller organisations — candidate screening, job matching, onboarding — is similarly exposed.
The third is the back office and traditional IT market. Large organisations have entire operational functions — finance, accounting, compliance, procurement, IT operations, regulatory reporting — that most small businesses cannot afford to staff adequately. AI is now making a credible version of that operational backbone accessible at a fraction of the cost.
As Index Ventures has noted, AI-native challengers in this space are entering existing services budgets rather than asking customers to create new software line items — which makes adoption structurally easier and the entry point lower. [9] In finance and operations specifically, AI-native startups now account for 91% of new tooling market share, because incumbent vendors face accuracy and compliance demands that slow their ability to ship AI-native workflows — creating exactly the kind of vacuum Christensen described. [9]
Enterprise spending on generative AI reached 37bn dollars in 2025, up from 11.5 bn in 2024. Vertical AI — solutions built for specific professional and operational domains — nearly tripled to 3.5bn dollars in the same year. [9] 80% of Y Combinator’s Winter 2025 cohort was AI-focused, with legal tech, financial services, healthcare support, recruiting and business operations among the most active application-layer categories. [6] Across the broader venture market, AI startups captured 52% of total global venture capital in 2025 — the first year on record where AI took more than half of all venture investment. [7]
AI handles the high-volume, lower-complexity layer of knowledge work and process based work that currently makes some services too expensive for most organisations — triage, document review, information retrieval, first-draft analysis, compliance checking, pattern recognition. It is not replacing professional judgment. It is making a credible lower-cost entry viable at the price point where excluded customers sit. A small team with domain expertise can now build in weeks what previously required months and significant capital. For most established organisations, their values will point their AI investment toward the existing business — which is precisely what leaves these markets open.
What an effective response looks like
The organisations that navigate this well act before the pressure to respond becomes unavoidable — because by the time it does, the challenger’s position is established. The most effective response is not defensive. The same analysis that identifies competitive vulnerability also maps where the largest underserved markets are and where new businesses can be built before external challengers establish a foothold. This means identifying which customer segments are being systematically underinvested in, designing propositions specifically for them, and incubating those efforts with the separation and investment criteria they need to develop on their own terms.
Four investment and governance principles determine whether that effort succeeds or fails. The first is applying different financial criteria — current margin expectations will make any early-stage disruptive opportunity look unattractive by design, and that is not a reason to walk away, it is a signal that the entry point is real. The second is genuine organisational separation, so the values of the existing business do not shape every decision the new effort makes. The third is enough time before measuring against incumbent financial standards — premature assessment is where most attempts at this fail. The fourth is treating these as genuine growth bets with real capital and real accountability, not as innovation programmes that signal ambition without committing resources to it. [8]
The established banks watched Monzo and Revolut build for a decade while investing in their own digital transformation. Western Union watched Wise and Remitly grow while continuing to optimise its agent network. In each case companies struggled to justify new investment decisions that ran against the grain of values correctly developed to defend the current position, and sustaining those decisions long enough for the new effort to find its footing.
AI compresses the timeline in which a challenger can establish a defensible position, across a wider range of industries, and from less predictable directions than previous cycles produced.
On the work this points toward
Christensen’s framework is often applied defensively — to understand what is coming before it is too late to respond. It is equally useful as a map for mapping new growth opportunities or so called “white spaces”. The same logic that identifies where incumbents are exposed points to where underserved markets are, where new competitive positions can be built, and where organisations applying different investment criteria can move to capture first mover advantage.
The organisations that will capture the next wave of growth and build competitive advantages will not just excel at responding fastest to AI enabled operational opportunities. They will tend to be the ones honest enough about where their values are pointing them — and deliberate enough to look seriously at the customers and markets those values make it easy to overlook.
Sources [1] Clayton Christensen, The Innovator’s Dilemma (Harvard Business Review Press, 1997) and The Innovator’s Solution (Harvard Business Review Press, 2003). [2] Revolut financial data: 2024 revenue, valuation and growth figures sourced from Revolut’s 2024 annual results and reporting in the Financial Times, November 2025. [3] Monzo customer figures: Monzo press release, 2024. https://monzo.com [4] Challenger market share data: QED Investors and BCG, Fintech’s Next Chapter, 2025. Nubank revenue: Business of Apps, 2025. Wise and Western Union growth comparison: FXC Intelligence, Digital Drives Money Transfers Q3 2025, November 2025. Stripe payment volume: Capital One Shopping Research, Stripe Statistics, 2025. [5] Anthropic, Labor Market Impacts of AI: A New Measure and Early Evidence, March 2026. [6] Y Combinator cohort data: CNBC, March 2025. YC fintech investment data: Growthlist YC Startups Guide, 2025. [7] Global VC AI investment data: BestBrokers, The State of AI Venture Capital in 2025, 2025. [8] Peter Hinssen, The Day After Tomorrow (Lannoo, 2017) and The Phoenix and the Unicorn (Lannoo, 2020). [9] Index Ventures, Automating the Back Office: How AI is Transforming Professional Services, September 2024. Menlo Ventures, 2025: The State of Generative AI in the Enterprise, December 2025.