A "Customer-First" IPO, Priced Like a Bank Nobody's Tested Yet
Nu Holdings Ltd. — the Cayman Islands holding company behind Nubank, the Brazilian digital bank — went public on the NYSE on December 9, 2021, in what was, at the time, the largest dually-listed IPO globally since 2012: a concurrent U.S. and Brazil listing (Class A ordinary shares on the NYSE, BDRs» on B3) that raised roughly US$2.8 billion. This is Nu's first annual report as a public company - its Form 20-F for the year ended December 31, 2021, filed with numbers that were already three weeks old by the time the stock started trading.
Nu's pitch, repeated by founder and CEO David Vélez on the earnings call, is that it's "a technology company that happens to be in financial services, and not a bank that has a better website or a better app" - proprietary core banking system, a low-cost operating platform, and an AI/machine-learning-first underwriting model applied to a Latin American consumer credit market the company's own investor materials describe as 55-65% unbanked across its three markets (Brazil, Mexico, Colombia). The IPO itself was staged as proof of that customer-first identity: more than 7.5 million customers were enrolled in "NuSócios," the company's own name for the incentive and reward program that gifted each of them a BDR, and over 800,000 customers made a paid reservation to buy more - the largest number of retail investors ever in a Brazilian IPO, by Nu's own account.
That "customer-first" framing sits awkwardly next to the actual voting arithmetic once the shares started trading (see Beyond the Usual below). It's a tension worth holding onto while reading the numbers: a business built on the story of financial inclusion, still controlled, vote for vote, by one person.
The Prescription
Nu's real edge, by its own numbers, is monthly cost to serve per active customer falling from $2.0 in 2018 to $0.8 in 2021 - a 60% reduction - while monthly ARPAM» recovered from $3.6 in 2020 to $4.5 in 2021 (still below 2018's $7.0, a mix effect of a much larger and younger customer base diluting the early, higher-spending cohorts), and the most mature customer cohorts already earn over $15/month. That gap between falling service cost and rising monetization per mature cohort is the entire bull case in one chart, and Nu should keep leaning into it precisely as management describes: more products per active customer (already averaging 3-4), not more marketing spend to acquire new ones. Customer acquisition cost of roughly $5 per customer - of which paid marketing is only about $1 - is the moat, and every dollar spent chasing volume instead of monetizing the existing 41.1 million monthly active customers dilutes it.
What Nu should stop doing: presenting Adjusted Net Income» as the headline profitability signal without equally emphasizing what's excluded to get there. The swing from a $26.8 million adjusted loss in 2020 to a $6.6 million adjusted profit in 2021 is real in the sense that the reconciliation is fully disclosed - but it exists because $225.4 million of share-based compensation (up from $56.3 million in 2020, a 4x jump) was stripped out, not because the underlying, audited net loss actually closed (it narrowed only 3.6%, from $171.5 million to $165.3 million). Titling an earnings-deck slide "Operating Leverage Starting to Show its Strength as We Break-Even on Adjusted Net Income," as Nu did, is a framing choice a skeptical reader should not accept at face value (see Beyond the Usual).
Key Financial Metrics
FY 2021 vs. FY 2020, consolidated, reported in USD - Nu reports natively in US dollars, so no FX conversion is needed
| Metric | FY 2021 | FY 2020 | YoY |
|---|---|---|---|
| Total revenue | $1,698.0M | $737.1M | ✅ +130% |
| Gross profit | $732.9M | $326.9M | ⚠️ +124% (margin: 43.2% vs. 44.3%) |
| Total operating expenses | $903.1M | $418.9M | ⚠️ +116% |
| Operating loss (gross profit less opex) | $(170.2)M | $(92.0)M | ⚠️ Loss widened |
| Loss before income taxes | $(170.2)M | $(193.2)M | ✅ Loss narrowed (2020 included a one-time $101.2M convertible-instrument charge) |
| Net loss for the year | $(165.3)M | $(171.5)M | ⚠️ Loss narrowed only 3.6% |
| Adjusted Net Income (Loss), non-IFRS | $6.6M | $(26.8)M | ⚠️ See caveat above and Beyond the Usual |
| Net cash used in operating activities | $(2,924.3)M | $974.5M | ⚠️ See note below |
| Total cash and cash equivalents (year-end) | $2,705.7M | $2,343.8M | ✅ +15.4% |
| Balance sheet metric | Dec 2021 | Dec 2020 | Change |
|---|---|---|---|
| Total assets | $19,858.7M | $10,154.3M | ✅ +95.6% |
| Total liabilities | $15,416.1M | $9,716.1M | ⚠️ +58.7% |
| Total equity | $4,442.5M | $438.1M | ✅ +914% (almost entirely IPO proceeds and preferred-share conversions, not retained earnings) |
| Deposits | $9,667.3M | $5,584.9M | ✅ +73.1% |
| Credit card receivables + loans to customers (net) | $5,975.3M | $3,083.6M | ✅ +93.8% |
Revenue growth of 130% (138% on an FX-neutral basis) was driven by both sides of the business at once: a 62% jump in total customers (to 53.9 million) and an 88% jump in monthly active customers (to 41.1 million, a 76% activity rate), compounding with rising monetization per customer. Gross profit grew slightly slower than revenue because ECL» provisioning - required upfront under IFRS 9 the moment a loan is originated, before the matching revenue has accrued - front-loads the cost of a rapidly growing credit book; credit loss allowance expense rose to 7.3% of the credit portfolio in 2021 from 5.1% in 2020, worth watching as Brazil's consumer-credit delinquency cycle normalizes back toward pre-pandemic levels (management's own stated base case on the call).
The operating cash flow swing - from generating $974.5 million in 2020 to consuming $2,924.3 million in 2021 - looks alarming in isolation but is a mechanical result of balance-sheet growth being classified as an operating activity under IFRS bank accounting: Nu built an entirely new $8.16 billion book of securities held at fair value through other comprehensive income (FVOCI») that didn't exist in 2020, on top of near-doubling its credit portfolio. None of that cash use reflects the core business burning money - it was funded by $2.6 billion of IPO proceeds, $800 million of preferred-share issuance completed earlier in 2021 ahead of the IPO, and continued deposit inflows, all captured in a $3.3 billion positive financing-activities line the operating and free-cash-flow figures above don't show.
The audited net loss barely moved. The only number that moved a lot was the one built by excluding $225 million of stock compensation from it.
Key Operational Metrics
- Customers: 53.9 million (Dec 2021) vs. 33.3 million (Dec 2020) - ✅ +62%. Monthly active customers: 41.1 million vs. 21.8 million - ✅ +88%, with activity rate improving from 66% to 76%.
- Purchase volume (credit/debit card spend, excludes PIX and wire transfers): $43.8 billion (FY2021) vs. $22.5 billion (FY2020) - ✅ +95% (+102% FX-neutral).
- Monthly ARPAM (average revenue per active customer): $4.5 (2021) vs. $3.6 (2020) - ✅ improving, though still a small fraction of the $35-38/month management says incumbent Brazilian banks earn per active customer - the stated multi-year monetization runway.
- Monthly cost to serve per active customer: $0.8 (2021) vs. $1.2 (2020) - ✅ -33%, the clearest evidence of the operating-leverage story management is selling.
- Geographic mix: Brazil generated $1,285.8 million of FY2021 revenue (97.5% of the total); Mexico contributed $29.5 million and Colombia $0.8 million - both still rounding errors on the P&L despite 1.4 million Mexican customers by year-end. Nu itself discloses only one reportable operating segment (its CEO, as Chief Operating Decision Maker, reviews the business on a single consolidated basis) - there's no separate country-level segment breakout to analyze beyond this revenue/asset disclosure.
- NPL 90+ (Brazil consumer finance, company data vs. market): Nu at 3.5% vs. a ~5.0% Brazilian market average at Q4 2021 - better than the market, though up from a pandemic-era low near 2.7% earlier in 2021, consistent with management's own "normalizing toward pre-COVID levels" framing on the call.
- Not available as a standalone figure in this filing: net interest margin or a CASA-style deposit-mix ratio in the form Indonesian/Indian bank filings disclose them - Nu's own credit and deposit metrics are reported on its own non-IFRS "FX Neutral," "ARPAM," and "cost to serve" framework rather than the NIM/CASA vocabulary this site's Indonesian-bank posts use (see, for a loose comparable, Bank Jago's FY2020 CASA and NIM disclosures - a much smaller, single-country neobank at a much earlier and less profitable stage than Nu was by this filing).
Beyond the Usual
One person controls three-quarters of the vote in a company that gave shares to millions of customers
Nu's Class A and Class B ordinary shares carry identical economic rights, but Class B shares carry 20 votes each against one vote for Class A. As of December 31, 2021, founder and CEO David Vélez Osorno - through Rua California Ltd., an entity he controls - held 992,000,922 Class B shares (86.2% of the Class B class), giving him 74.9% of Nu's total voting power while the IPO simultaneously handed a symbolic BDR to more than 7.5 million ordinary customers under the NuSócios program. Public Class A shareholders, including every one of those millions of gifted-BDR customers, have essentially no ability to influence a board vote, a related-party transaction, or an executive-pay decision regardless of how many of them agree. This is a common structure among newly public founder-led tech companies, not evidence of wrongdoing - but it's worth reading against the IPO's own "customer-first," "challenging the status quo" marketing before assuming the offering meaningfully democratized control of the company, rather than just its economics.
A board member's company got a five-year, R$36 million marketing contract
On June 30, 2021, Nu signed a five-year marketing and publicity agreement worth a total of R$35,950,617 with Rodamoinho Produtora de Eventos Ltda., a company controlled by Larissa de Macedo Machado - the Brazilian singer known as Anitta, and a sitting member of Nu's board of directors. Part of the payment is being satisfied through RSU» issuance rather than cash. The deal was signed a few months before Nu adopted a formal related-party transaction policy in October 2021 requiring board or committee approval for such arrangements going forward - meaning this specific contract predates the governance process that would now review it.
Adjusted Net Income turned positive almost entirely because $225 million of stock compensation was excluded from it
Nu's own non-IFRS reconciliation shows exactly how the swing from a $26.8 million Adjusted Net Loss (2020) to a $6.6 million Adjusted Net Income (2021) happened: start from the $165.0 million loss attributable to shareholders, add back $225.4 million of share-based compensation, subtract an allocated $60.0 million tax effect on that add-back, add back $11.2 million of Customer Program (NuSócios) expense, and subtract a further $5.0 million allocated tax effect on that item - and the number crosses from deeply negative to modestly positive. Share-based compensation itself isn't a one-time or made-up cost - it dilutes existing shareholders exactly like cash compensation would, just without a cash outflow - and the same table shows it nearly quadrupled year over year (from $56.3 million to $225.4 million), which the earnings materials don't put nearly as much emphasis on as the "Adjusted Net Income" figure it makes possible.
A $422.6 million CEO award, tied to the stock and pledged to charity
Nu's board granted David Vélez a set of "2021 Contingent Share Awards" on November 22, 2021 - Class A shares equal to up to 2% of fully-diluted shares in issue, vesting only if Nu's share price closes above $18.69 and then $35.30 for 60 consecutive trading days (roughly 2.1x and 3.9x the IPO price of $9.00), with a five-year minimum service requirement layered on top. The total award was valued at $422.6 million, to be expensed as compensation over 7.5 years. Vélez has committed to donate all shares resulting from the award to his family's philanthropic platform, and signed The Giving Pledge in August 2021. Neither milestone had been achieved as of year-end.
A regulated payment institution, not a licensed bank, sits at the center of the group
Nu's core Brazilian consumer entity, Nu Pagamentos, operates under a Central Bank of Brazil "payment institution" authorization - not a full banking license. Nu Financeira (which separately committed to the Central Bank to hold a 14.0% Basel capital adequacy ratio through 2023, above what's required of most Brazilian banks), Nu DTVM, and NuInvest hold the group's actual financial-institution licenses. It's a structural nuance worth knowing before assuming "Nubank" operates as a single chartered bank the way BCA or Bank Jago do in Indonesia.
A debit-card interchange lawsuit management calls "remote," and routine Brazilian tax litigation
On October 14, 2021, Getnet Adquirência e Serviços para Meios de Pagamento S.A. sued Nu Pagamentos and Mastercard Brasil, seeking to cap the interchange fees Nu charges on transactions run through Getnet's point-of-sale devices, alleging roughly R$64 million in damages from the higher fees currently charged. A Brazilian court denied Getnet's preliminary injunction request in October 2021, and Nu itself estimates the likelihood of loss as remote. Separately, Nu's total provision for legal proceedings was $18.1 million at year-end (with $17.5 million in matching judicial deposits), of which $17.1 million is tax-risk provisions - $14.9 million of that specifically an ongoing PIS/COFINS tax dispute where Nu is awaiting release of its judicial deposits, expected around December 2023 - unremarkable by the standards of Brazilian corporate litigation, but worth having on record as this backfill continues.
What Management Emphasized on Nu's First Earnings Call
The call - Nu's first as a public company - opened with several minutes of Vélez narrating the IPO itself before a single quarterly number was discussed: the dual U.S./Brazil listing structure, the NuSócios BDR giveaway, and the "four pillars" (mission-driven culture, customer obsession, proprietary technology, AI/ML-first approach) framed explicitly as differentiators rather than aspirations. That's the strategic framing worth reading before the numbers below it - management's stated theory of the business, not a recap of results.
On the numbers, CFO Guilherme Lago told the call that "as a result of our growing scale, we are beginning to reap the benefits of operating leverage on an adjusted net income basis" - directly under a slide titled "Operating Leverage Starting to Show its Strength as We Break-Even on Adjusted Net Income" - language that leans on the Adjusted Net Income figure covered in Beyond the Usual above without volunteering, unprompted, how much of that swing came from excluding a near-quadrupled stock-compensation charge. When an HSBC analyst asked directly about rising cost-of-risk expectations for 2022, COO Youssef Lahrech gave a real, specific answer (delinquency normalizing toward pre-COVID levels, credit mix shifting toward higher-margin/higher-risk personal loans) rather than deflecting - management was substantially more forthcoming on credit-quality trends than on the framing of its own headline profitability metric.
Three topics that came up nowhere in the prepared remarks or the full Q&A session: the Getnet interchange lawsuit, the dual-class voting structure that leaves Vélez controlling three-quarters of the vote, and the $422.6 million contingent share award granted to Vélez about two and a half weeks before the IPO. None of these are concealed - all three are disclosed plainly in the 20-F itself - but a listener relying only on the call would have no reason to know any of them exist.
Target Valuation Range
No numeric fair-value target yet - only one quarter of public trading exists (three weeks of price history, one audited fiscal year), so the ~24x EV/Revenue and ~9.7x P/B below describe today's ~$40.7 billion enterprise value, not a derived target. A DCF or reverse-DCF would manufacture false precision on this little data; the honest verdict is that this is expensively priced for a company that is still IFRS-loss-making, on the strength of a growth story and a non-IFRS profitability inflection that owes most of its size to excluding stock compensation. Nothing here proves the growth won't eventually justify the price - but nothing in FY2021's audited numbers proves it yet either.
Nu closed out its debut three weeks of trading at $9.38 on December 31, 2021 - down 9.2% from its $10.33 opening-day close and down 20.8% from its $11.85 closing high the following session, though still above its $9.00 IPO price. With 3,459,743,432 Class A and 1,150,245,114 Class B ordinary shares outstanding at year-end (before a January 2022 over-allotment exercise that added a further 27.6 million Class A shares as a subsequent event):
| Market cap → enterprise value | FY2021 |
|---|---|
| Share price (period-end) | $9.38 |
| Shares outstanding | 4,609,988,546 |
| Market capitalization | ~$43.2B |
| Total liabilities | n/a (net cash basis used) |
| Less: cash and equivalents | $2,705.7M (vs. $157.3M interest-bearing debt) |
| Enterprise value | ~$40.7B |
| Peer-multiple sanity check | FY2021 |
|---|---|
| Revenue | n/a |
| Enterprise value | ~$40.7B |
| EV/Revenue | ~24x - rich by any standard, especially set against an IFRS net loss and gross margin that compressed slightly year over year (44.3% to 43.2%) |
| P/B» | ~9.7x book value per share of $0.96 (total equity of $4,442.5 million ÷ 4,609,988,546 shares) - far below Bank Jago's roughly 31x book at its own FY2020 backfilled quarter, but still a premium multiple for a company posting a negative return on equity (approximately -6.8% for FY2021) |
| P/E | not meaningful - Nu posted an IFRS net loss for the year |
This is Nu's first quarter as a public company, so no prior-quarter comparison exists yet. A full DCF isn't included here for the same reason it wasn't appropriate for Bank Jago's own first backfilled quarter: one year of public-company numbers, with revenue still growing well over 100% annually and margins not yet stabilized, doesn't support a credible multi-year free-cash-flow projection - a precise-looking DCF built on this little data would be false confidence, not real analysis. The peer-multiple read above is the honest lens for this quarter: the market is pricing in years of the ARPAM-expansion and cost-to-serve story management is telling, and this filing's own numbers - a widening operating loss, provisioning pressure from rapid credit growth, and a profitability "inflection" that leans heavily on excluding real compensation costs - don't yet confirm or refute whether that story plays out. A formal two-year price-history comparison isn't possible for this post either: Nu has only traded publicly since December 9, 2021, giving barely three weeks of price history as of this quarter's end, mirroring the same constraint noted in Grab's own debut-quarter post.
Nu Holdings Ltd.'s Annual Report on Form 20-F for the fiscal year ended December 31, 2021 (including audited consolidated financial statements and notes), its Fourth Quarter 2021 Results earnings presentation, and its Q4'21 earnings conference call transcript (presentation and call both dated February 22, 2022).