In this article
Nubank hired more slowly than planned in the third quarter of 2026, its finance chief Rob Livingston told Reuters on November 12. Livingston said AI was helping the Brazilian digital bank increase capacity and productivity, allowing it to slow recruitment. Parent company Nu Holdings now expects its full-year efficiency ratio to come in at 19% to 19.5%, down from a previous estimate of around 20%. The account describes a change in the pace of hiring. It gives no figures for layoffs or changes in total headcount. Reuters reporting via CNA
What slower hiring tells us
Livingston compared recruitment with what the company had expected. An operation that has planned to hire more people to meet growing demand can revise that plan if its existing teams can handle more work. It may still add employees. The size of the workforce at the end of the period depends on how many people joined and how many left, including through resignations and other departures.
The interview did not identify how many planned positions went unfilled or say which teams slowed recruitment. To establish what happened to the workforce, a reader would need headcount figures measured on a consistent basis, as well as hires and departures over the same period. That evidence would also help distinguish a change concentrated in one function from a pattern across the bank. None of those questions can be answered by the change in the efficiency ratio.
Two revenue figures in the same earnings presentation
Nu Holdings reported about $6.26 billion in gross revenue on its managerial basis for the three months through September. The consolidated accounts, prepared under International Financial Reporting Standards, show $5.82 billion in total revenue and $1.09 billion in net income. The earnings presentation labels the two revenue measures separately. Comparisons across quarters should use the same reporting basis.
Revenue records what the business earns from its operations, including interest and fees under the applicable definition. Net income reflects what remains after costs, expenses, taxes and other results are accounted for. At a lender, the cost of attracting deposits and the expense of expected credit losses can materially affect what reaches the bottom line. Revenue can rise while either of those costs absorbs some of the increase. Nu Holdings earnings presentation filed with the SEC
Why the ratio can fall while spending rises
Nu divides total operating expenses by a revenue measure that combines net interest income and fee income after the specified transaction costs and revenue-based taxes. For the third quarter, the presentation rounds operating expenses to $846 million and the revenue base to $4.386 billion. A ratio of 19.3% means roughly $19.30 in operating expenses for every $100 of revenue on that definition.
Nu reported 19.5% in the previous quarter and 20.3% in the same quarter a year earlier. The year-over-year difference was one percentage point. The latest ratio fell even as operating expenses rose because the denominator grew faster. The new range of 19% to 19.5% is management's forecast for the full year.
The earnings figures combine several changes
The company's release cited lower personnel costs and some expenses moving into the fourth quarter as factors in the quarterly result. Livingston told Reuters that stronger-than-expected revenue also helped the outlook. He said the rollout of the return-to-office policy was moving more slowly than planned while the company opened offices in several cities. These developments make it hard to read the change in the ratio as a measure of AI's contribution.
When an expense shifts into the next quarter, the effect on the current figures may be temporary. When revenue exceeds expectations, the ratio can improve before any change in the staffing plan is visible. To estimate AI's contribution, an analyst would need to compare similar work with and without the tools and account for changes in demand, product mix and implementation costs. The earnings figures describe the combined outcome at the bank.
Where teams could gain capacity
A financial operations team might use a tool to extract information from documents so a reviewer can spend more time on exceptions. In software development, a system might draft a test that an engineer checks against what a service is supposed to do. Those are examples of possible uses in banking. The interview does not identify which of them, if any, led Nu to revise its hiring plans.
Whether such tools increase capacity depends on the whole task. A first draft that takes seconds still needs review. A document tool that sends a routine case to the right team can save time, while a mistake on a complex case may create more work downstream. A useful comparison would measure how long cases take to resolve and whether the error rate changes. Teams handling financial information would also need to check what data the tool can access and who can review its outputs.
What to watch in hiring data
Entry-level work gives new employees a chance to learn how a team makes decisions. If a system handles more of the initial work, employers may need to teach those skills differently. A junior engineer reviewing a generated test still has to understand which failure matters to the customer. A new operations employee checking a summary still needs to know when a missing detail changes the outcome.
Nubank's account gives one example of management changing its hiring plans after judging that existing teams could handle more. Another employer could use extra capacity to expand a service or manage rising demand. To see whether the pattern spreads through banking, compare hiring and headcount over several quarters by company and job function. It would also help to know whether employers are filling fewer entry-level posts, moving work between teams or changing who handles exceptions.
For the wider labor market, Michael Barr discussed jobs and inequality in the United States.
