Nu Holdings Lost $16.4 Billion in One Pre-Market Session — and the Rate Everyone Blamed Is Not the Risk

Luiz Inácio Lula da Silva won Brazil's presidential runoff on Sunday, Oct. 25, 2026, taking 51.8% of the valid vote against Flávio Bolsonaro's 48.2%. The market had priced the opposite: after Bolsonaro led the first round on Oct. 4, the Ibovespa closed almost 7% above its pre-vote level, the real strengthened nearly 5% against the dollar, and Brazil's five-year credit default swap spread fell 22 basis points to 108 basis points, a six-year low. By Monday's pre-market, that fiscal-adjustment trade was being unwound in a single session.

The overnight tape sorted Brazilian equities by one variable, and it was not fintech. Every Brazil-domestic consumer name fell in the 16% to 23% range, while the dollar-linked exporter barely moved: Vale lost 3.6% against a 19.6% fall in the broad Brazil ETF. The regional consumer platform MercadoLibre fell 10.9%, less than half of Nu's 22.5%. Nu's extra 12 points of downside measured Brazil concentration rather than fintech risk, because 118 million of its 139 million customers and $36.4 billion of its $45.3 billion in deposits sit in Brazil.

The reflex reading of a Lula win is \"no more rate cuts,\" and on the surface the tape confirms it: Brazil's central bank has held the Selic at 13.75% since Sept. 16, 2026 after five cuts this year, and economists were forecasting 13.50% by year-end before the vote. Nu's own numbers say the policy rate is the least of its problem. Its deposit base costs 88% of the interbank rate, roughly 12% at a 13.75% Selic, and management attributed the second-quarter margin expansion to credit income and a lower cost of credit, with float income and funding costs roughly neutral.

That table is the whole argument. Nu's margin was built on credit growth and falling credit cost, not on falling rates, so a central bank that holds at 13.75% through year-end costs Nu far less than the market assumed. What the result actually threatens is the denominator of the story: the credit cycle, the real, and the multiple.

Nu's balance sheet is more exposed to Brazilian sovereign risk than any headline suggests. At June 30, 2026 it carried $30.5 billion of investments, of which $13.67 billion was Latin American government bonds, against total shareholders' equity of $13.25 billion. A fiscal-driven repricing of the long end therefore lands on a book more than twice the size of the equity behind it, and Nu's float income is earned on exactly that book.

The channel that should worry investors most was already deteriorating quietly before the election. Nu's 90-plus-day delinquency ratio rose 35 basis points to 6.9% in the second quarter while the earlier-stage 15-to-90-day ratio improved 16 basis points to 4.8% — a rotation management attributed to seasonal migration of first-quarter delinquencies. The allowance for credit losses grew to $6.6 billion, covering 90-plus-day nonperforming loans 244%. Cost of credit fell 9% sequentially to $1.69 billion, and management said the Desenrola household debt-relief programme cut cost of credit by about 5%, with more of that effect expected in the third quarter.

Here is the paradox. A fiscal-expansionary government is, in the short run, a credit-cost-relief beneficiary for Brazilian lenders: programmes like Desenrola let borrowers renegotiate rather than default, which flatters provisioning while capping the interest Nu can charge. Relief lowers the charge-off rate today and erodes the revenue base tomorrow. With growth expected to slow to about 1.3% a year from 2.7% in the current term, and a deficit above 9% of GDP, the demand side of the book is what deteriorates first.

Nu is not exposed to Brazil in one place; it is exposed along a chain, and the election hit each link differently. Upstream, the price of money, the float assets and the card rails all sit in or under Brazil's sovereign risk. Downstream, the $39.4 billion credit book absorbs Brazilian household and small-business balance sheets, and the same rate shock lands on every Brazilian payments and credit platform.

Nu launched its United States business on Sept. 10, 2026: a savings account paying 3.50% APY with deposits held at the FDIC-insured Lead Bank, a no-annual-fee Mastercard credit card with 1.5% cashback, and Nu Global accounts paying 3.50% on USDC balances and 2.20% on EURC with a virtual card and transfers across more than 35 countries. The rate comparison is what matters now: Nu's Brazilian deposits cost 88% of the interbank rate, roughly 12%, while its dollar deposits cost 3.50%. For the first time Nu holds a funding source that Brazil's rate committee cannot reprice, which is why the September launch matters more after this election than before it.

At the Oct. 23 close of $15.16, Nu traded at 20.5 times trailing earnings and 5.56 times book, on a 30.3% trailing return on equity. Marked to the $11.75 pre-market level, those multiples fall to roughly 15.9 times and 4.31 times. Two adjustments matter. A 10% weaker real cuts dollar-translated earnings by about 9%, so the honest post-drop multiple is closer to 17.5 times, not 15.9 — roughly a third of the headline de-rating is currency translation of the same earnings, and the rest is the multiple.

The comparison says two things. Nu still carries the highest return on equity in the group and, after the drop, a multiple that is no longer a premium to XP Inc, a platform with no Brazilian consumer credit book at all. And Banco Bradesco at 0.98 times book on a 14.2% return on equity is priced for a credit cycle that normalises — a 1.3% growth path with a 13.75% policy rate does not normalise, it stalls. The board's $1.0 billion repurchase authorisation, running from June 4, 2026 through June 3, 2027, buys more stock at $11.75 than at $15.16, though it was sized when the shares were near their 52-week high of $18.98.

Over days to quarters, three dated things matter: third-quarter results on Nov. 12, 2026, the remaining 2026 rate decisions, and the shape of the 2027 budget. Capital Economics expects the Selic to stay at 13.75% to the end of 2026, and a survey of 73 investors cited by Citigroup found most expected the currency to fall at least 5% and swap rates to rise by more than a percentage point — a floor under the cost of money, not a ceiling on it. The new administration's stated ambition is a budget surplus of about 1.3% of GDP by 2030.

Over one to three years, the case for owning Nu has little to do with Brazil's president and a lot to do with whether Nu becomes a multi-country bank. Mexico is the proof of concept: 16 million customers by the end of July 2026, a banking licence granted and a transition from credit-first fintech to full digital bank, ARPAC of 12.3 against 5.6 in Brazil, and break-even in six years versus eight in Brazil. Mexico held $5.7 billion of the $45.3 billion deposit base at June 30, 2026, so the diversification is real but early. The United States is earlier still, with a 12-to-30-month model build before a single loan is written. Brazil's politics sets the next twelve months of Nu's multiple; Mexico and the dollar book set the next five years of its earnings.

The bull case is that a 22% haircut took a lender with a 33% quarterly return on equity, a 12.4% risk-adjusted margin and 244% coverage of 90-plus-day loans down to 15.9 times trailing earnings, while the thing that broke — a fiscal adjustment priced into the curve — is reversible with a credible 2027 budget. The bear case is that a 10% weaker real cuts dollar earnings, 1.3% growth stalls the credit cycle, the 6.9% delinquency rate has further to run, and a sovereign securities book 2.3 times the size of the equity is marked against a curve that has not yet moved.