Competition, Intensity & Building a $115 Billion Global Bank | Nik Storonsky, Revolut
Summary
- Storonsky wants Revolut to become the number one bank in at least 100 countries, using shared technology to undercut fragmented incumbents. He puts its footprint at “probably thirty-eight” countries, adding 10–15 markets every one to two years. One configurable platform and ledger support lower operating costs across countries; “I’m coming for his dinner” extends the ambition beyond retail into enterprise and potentially investment banking.
- Bank licensing is central to Storonsky’s mass-market adoption thesis, and postponing it created a costly sequencing problem. Faster e-money licensing brought millions of customers, but applying for a UK banking license with seven or eight million customers made regulatory approval harder; the process took almost five years. He reports 80% adult penetration in Ireland and about 20% in the UK and France, arguing that reaching mass adoption requires customer trust: “you need to be a bank.”
- Revolut’s product expansion follows portfolio rules designed to override intuition. Teams generally capped at ten build products over nine to 18 months; bottom-quartile bets are sunset, while up to the top 25% receive more teams and resources. After roughly 50 bets over ten years, Storonsky says, “I don’t really trust my intuition” about which products will work.
- AI is already changing operating costs, while smaller product teams remain an organizational redesign in progress. Storonsky says 90% of customer support is automated, following a support organization of more than 5,000 people, and reports a 30% increase in engineering speed. He thinks product teams should shrink to two or three people and is betting on a “full stack AI employee” with calls, email, Slack and internal-system access.
- European constraints shaped an efficiency culture that combines tight resource allocation with demanding personnel decisions. Storonsky says Revolut moved from negative gross margins lasting “three, four, five years” to as much as 80% positive gross margin. He argues that abundant funding encourages founders to solve problems with money before developing repeatable systems, and explicitly prioritizes winning over retaining average performers: “likable is not the goal.”
- Storonsky treats investor control as an operating risk and has built a competing investment business around statistical selection. He says Revolut now has no VCs on its board and compares investors directing operations to “cheerleaders” trying to play the match. Separately, Quantum Light’s fully invested $250 million Series B fund has a reported 56% IRR; a $500 million Series C fund is starting to invest.
- Physical advertising serves both customer trust and an attempt to buy impressions below digital prices. After seven years of organic growth, Revolut recruited marketing expertise from gaming and expanded into airport advertising and a Barcelona experience store. Storonsky says ads appear in probably 50% of European airports; Senra’s description of a “cocoon of Revolut ads” captures the intended visibility.
- The eventual destination for banking cash flow includes technology bets beyond financial services. Revolut has built Pragma, a transformer model trained on transaction data, with 30 million transactions generated daily. Storonsky would like to fund deeper model development and data centers, but only hopes Revolut might have sufficient resources in three to five years; those moonshots “might not necessarily produce P&L” for another three to five years.
Deep dive
1. Revolut’s target is first place in 100 countries
Storonsky’s ambition starts with being “number one in at least a hundred countries.” Revolut began roughly twelve years earlier with a card offering interbank foreign-exchange rates, then added credit, investing, crypto, business accounts and nonfinancial services. It now operates in “probably thirty-eight” countries, with banking licenses in the majority, and is adding ten or fifteen markets every one to two years.
Senra’s recollection of “I’m coming for his dinner,” directed at Jamie Dimon’s business, gets a concrete explanation: retail is only part of the opportunity. Storonsky also wants enterprise business and potentially investment banking, which he describes as manual and expensive. His competitive discipline is inward: “I don’t really look at competition.” Instead, he asks how he would outcompete himself and demands “almost hundred percent every day.”
His claimed opening is incentive alignment: professional CEOs stay “three, four years,” collect bonuses and defend their jobs; he regards himself and Revolut as “one entity.” Founder-led incumbents, he argues, would already have better products. Against regulators’ concern about excessive risk-taking, his answer is that dependence on the company’s survival discourages “stupid excessive risk.” Senra agrees that founders cannot cleanly separate business from personal identity.
2. Licensing becomes harder after customers arrive
The initial choice favored speed: an e-money license could take one year, against three or four for a banking license. But applying after acquiring millions of customers changed the regulator’s incentives. A failed startup with no customers creates little exposure; a failed bank with millions invites blame. Revolut applied with seven or eight million UK customers, Storonsky says, and spent almost five years obtaining the license.
Senra repeatedly presses why a fintech cannot become very large without banking status. Storonsky’s answer is trust: customers must feel comfortable depositing salaries and wealth, then using investment products. He says he thinks Chime and Cash App made a big mistake by avoiding licenses; he argues unlicensed firms can target niche cohorts, including low-income customers, but cannot reach prime customers or become a prime bank account. Early adopters and their networks might produce 4% penetration, while his ambition is 40%, 50%, 60% or 70%.
Ireland supplies his strongest example: “Four out of five adults use Revolut accounts.” He attributes that 80% penetration to four weak incumbent banks, a small country and many people living outside Ireland as well. He puts the UK and France at about 20%, with certain other markets at 30–40%. Ireland demonstrates what he thinks is possible elsewhere, conditional on becoming a bank.
Legacy technology reinforces the opening. Replacing an old third-party core requires customer migration and introduces execution risk, so risk-averse managers avoid it. Storonsky argues that weak technical understanding compounds that reluctance: “doing anything is risky” when management lacks the skills. Banking requires building technology alongside regulation and compliance, and he considers incumbent management poor at that work.
3. Product bets survive on measured growth
Storonsky describes organization building as repeatedly repairing a machine’s bottlenecks. When headcount growth reduced speed, he created small, independent teams sharing a platform. Regulation then became the constraint, prompting regulatory and government-affairs capabilities; hiring became another, requiring recruitment processes. “You build something, then you hit a bottleneck,” identify it, solve it and repeat. The machine’s output is a product people use.
His trading background contributed risk management and quantification. “Metrics and KPIs” kill internal politics, he argues, because individual contributors and managers face visible goals and objective outcomes. The product-bet process added a missing discipline: Revolut had been releasing many features without adequately tracking performance. Proposals can originate from employees or leadership, but receive scrutiny before a team is funded.
A funded bet generally receives up to ten people across engineering, design, product and sometimes operations or data science, then takes nine to 18 months to launch. Gross-profit growth determines what happens next. Across roughly 50 bets over ten years, he says the best have grown 10–20% month on month, illustrating their progression with one million, two million and three million of monthly gross profit.
A new product landing in the historical bottom 25% is sunset, with weak bets cut within 18 months. Up to the top 25% receive additional resources: one team becomes several, then a department and a major P&L line. The discipline applies to his own ideas: “I don’t really trust my intuition,” because he has been wrong repeatedly. His preferred approach is “completely algorithmic.”
4. AI changes the size and capabilities of a team
Storonsky reports that 90% of customer support is automated; the function previously employed more than 5,000 people and now employs substantially fewer. Engineering speed has increased 30%, while LLMs also handle translation and cross-selling. He nevertheless considers this “just in the beginning.” The implication extends beyond making the existing organization faster: its structure needs to be reconsidered.
The ten-person product team “should be two people, three people at best,” he argues, because one developer can cover front end, back end and even data science. His larger bet is a “full stack AI employee” that joins calls, has email and Slack accounts, accesses internal back-office systems and performs work. He thinks this is “completely doable” with current technology.
His concrete example of spreading innovation comes from design. One team supplied an LLM with Revolut’s design system at code level and prompted it to generate flows. The output was a working, clickable interface on a laptop or simulator, produced three or four times faster than Figma screens. Storonsky is pushing that approach across designers, consistent with his insistence that useful innovations become standard equipment throughout the company.
5. Autonomous people make a wide reporting structure possible
Storonsky has about 50–52 direct reports and also reaches individual designers, product owners and developers on important projects. His criticism of some managers is specific: they merely collect information and pass it to stakeholders without improving direction, introducing innovation or finding an easier route. His preferred “self-guided missiles” identify their own goals and reach them, minimizing the management time required.
His hierarchy distinguishes excellent people who identify goals, good people who need a goal supplied, and average people who need repeated guidance. Below-average people may still miss the goal despite that extra time. Asked whether he fires average performers, he answers, “I do, yes.” Winning takes priority over empathy in these decisions. He allows that strong executives might be liked, but “likable is not the goal”; achieving the objective with minimal resources is.
His counterexample to advice about hiring experienced executives is blunt: probably 50-plus such hires in Revolut’s first five years, only two retained. Past achievement may cease to predict future fit as priorities, capabilities and competition change. He instead looks for career progression in the top 1% of peers, seeking people around 25 or 30 who can continue that trajectory inside Revolut.
6. Constraints teach skills that abundant capital can conceal
Storonsky sees California’s weakness as a culture of solving problems with money. His example is a B2B company reaching one million ARR, then hiring an expensive VP of Sales who recruits 50 salespeople. First build a cheap, repeatable selling system with attractive payback, he argues; then add employees. Otherwise, “you’re not really developing yourself as entrepreneur,” and become vulnerable when funding disappears.
Revolut’s own economics initially offered little room for waste. Storonsky says gross margins stayed negative for “three, four, five years” before reaching as much as 80% positive gross margin. He contrasts European interchange of “twenty base points” with “one half percent” in the US, alongside European restrictions on credit. His explanation for survival is sustained effort to optimize individual problems rather than spending through them.
A European banking license also did not automatically deliver local primary-account functionality. Each country required a local branch, office, connectivity and regulatory work, plus language differences. Tough employment laws demanded care in hiring. He thinks building through those constraints taught Revolut capabilities that would make certain aspects of expansion into the US, Brazil or Mexico easier, where he considers credit and interchange economics more favorable.
Senra challenges the claim that America is not especially competitive. Storonsky says Europe is even less competitive and points to China’s 996 schedule and technical education. His weekend-work argument rests on personal experience: of 140 companies in his original coworking space, three survived five years and two survived ten; all five-year survivors worked weekends. He says most of his direct reports also work his long hours.
7. Investor control can become an operating hazard
Storonsky’s “cheerleaders” analogy concerns investors directing a game they have not played professionally. Early European fundraising involved liquidation preferences, board seats and reduced founder voting rights. He considers raising LP capital on “two and 20” economics much easier than building an operating business, and argues that board observation does not teach execution. The danger arises when those investors use contractual power to impose incorrect operating advice.
His sharpest example is an investor demanding that he fire his best person for being “too hardcore.” Storonsky refused; even a coalition of investors eventually backed down. He says successive financing rounds removed VC influence, leaving no VCs on Revolut’s board. He prefers operators who have built businesses and urges founders to negotiate agreements keeping investors purely financial, without decision-making powers.
8. Quantum Light applies statistical selection to venture investing
Quantum Light originated, he thinks, in 2022 or 2023 after conflict with a major investor that entered Revolut at a $33 billion valuation and held rights blocking shareholder-agreement changes. Frustration became competitive motivation: “I also need to compete with VCs in their business.” Its model selects 30–40 companies over two or three years, investing equal amounts. His ambition is something like BlackRock in private markets: “we’ll see” whether it gets there.
The first fund was $250 million, focused on Series B, and is fully invested; Storonsky reports 56% IRR. The second is $500 million and is starting to invest in Series C companies. He recalls supplying 20% or 25% of the first fund himself, with other founders supplying the rest; the second includes institutional LPs. He thinks returns rank in the top 5% of funds.
Model inputs include investing funds, individual partners, their performance and age; he says the best partner cohorts are 30–40, while stressing age is only one variable. Support adds Revolut’s hiring, performance-management and expansion playbooks, his advice, and potential access to Revolut as an enterprise customer or to its users. Quantum Light is separate from Revolut and run by someone else.
9. Physical visibility helps a branchless bank earn trust
Marketing was organic for the first seven years. Recalling 2021, Storonsky says revenue fell, “I think seventy percent,” as COVID eliminated travel and exposed dependence on the travel-card business. When he turned seriously to marketing, he sought expertise from gaming companies with billion-plus budgets and little margin for error. The resulting machine focused on performance advertising, lifetime value and payback; the first marketing hire was fired.
A banking license sharpened the trust problem: without branches, customers cannot physically see Revolut as they see HSBC, Lloyds or JPMorgan Chase. Storonsky also saw an impression-price opportunity. He says Revolut now advertises in probably 50% of European airports, calculating travel frequency, impressions and cost against digital alternatives. His explanation of the immersive placements is simple: “Just buying full airport.”
Senra pushes beyond the cost-per-impression explanation: entire hallways create ubiquity, a “cocoon of Revolut ads,” rather than a fleeting sign. Storonsky agrees that trust matters too. The first Barcelona Revolut experience store follows similar logic: an Apple-like setting to try products and generate impressions among visitors and passersby. He evaluates its physical visibility against the price of reaching people digitally.
10. Global banking cash flow is intended to finance deeper technology
Asked why an American should choose Revolut over JPMorgan, Storonsky pitches domestic functionality plus global accounts, instant international transfers and a cheaper, better interface. US customers would be able to open accounts in Europe or Australia. Revenue leans more toward subscriptions, transactions and commissions than credit. Revolut offers cards and loans, but has not sold that credit off because it remains a small balance-sheet component.
The scaling mechanism is “one system, one ledger” across countries. Legacy banks expanded through acquisitions and accumulated incompatible systems; extracting consolidated reports can take months, he says. Revolut changes configurations for onboarding in the UK, US or Mexico instead of rebuilding each bank. Lower maintenance staffing and reusable product and marketing knowledge produce economies of scale. With time, he thinks this can become “an unstoppable machine.”
Pragma is an existing move into deeper technology: a foundational transformer model trained on transactions rather than words. Storonsky says Revolut processes 30 million transactions daily, giving it billions of training examples. Potential applications include fraud detection, credit and possibly investing. His broader aspiration includes building data centers and training additional models, but those ambitions require substantially more investment and cash flow.
The eventual moonshots would “most likely” extend beyond financial services. He calls the current business “the same old banking but done in a new way”: cheaper, better products connected inside one app. He wants eventually to build things never done before in technology. Those ideas remain in his head; he hopes Revolut might have sufficient resources in three to five years for bets that “might not necessarily produce P&L” for three to five years.