NU Nu Just Booked Its First $1 Billion Profit Quarter. Its Cost Base Grew Faster Than Revenue To Get There.

A Billion-Dollar Quarter, Bought With a Worse Efficiency Ratio

Nu Holdings' second quarter of 2026 (period ended June 30, 2026) is the quarter the company's own founder has been building toward since the 2013 pitch: net income for the period reached $1,061.1 million, the first time Nu has crossed the billion-dollar mark in a single quarter, up 66.6% from $637.0 million a year earlier on an IFRS basis (management's own FX-neutral framing puts it at +49%). Customers reached 138.9 million globally, ROE closed at 33%, and - as covered in nu/2026-03 - Mexico's break-even milestone from the prior quarter has now compounded into an actual banking license: Nubank Mexico received its final Operations Authorization from Mexico's banking regulator (CNBV) on July 9, 2026 and began operating as a full multiple bank on August 6, 2026, becoming, in the company's own words, "the largest digital bank in the country."

That's the headline. The tension a reader following the numbers rather than the press release should notice is what happened underneath the profit record: total operating expenses grew 65.6% year-over-year, outrunning the 50.3% growth in total revenue that funded them. The company's own Efficiency Ratio^» - the ratio it uses to tell investors how disciplined its cost base is - worsened to 19.5% this quarter from 17.6% the quarter before, and management's own explanation is candid about why: "real estate and marketing expenses shifted from the first quarter into the second," on top of "continued investments in international expansion." A billion-dollar profit quarter built partly on real estate and marketing timing, not a step-change in operating leverage, is a different quarter than the headline number alone suggests.

The Prescription

Nu should keep pushing the Mexico playbook exactly the way it's now proven it works: ARPAC^» there is already running ahead of Brazil's equivalent early-cohort pace ($12.3 vs. $5.6 "at the same stage," per the company's own disclosure), and a full banking license removes the last structural constraint on deposit-funded growth in that market. The Banco Porto Real acquisition announced July 20, 2026 (see Beyond the Usual) - adding a formal banking license to the Brazilian conglomerate - is the same instinct applied at home: close regulatory gaps before a rule change forces the issue, rather than after.

What Nu should stop doing: treating this quarter's Efficiency Ratio deterioration as a footnote to a profit-record headline. A bank that leads its investor materials with "more than a billion dollars in quarterly net income" while its own cost discipline metric moves the wrong way for the second time in three quarters (19.9% in Q4'25, 17.6% in Q1'26, now back up to 19.5%) is letting a genuinely strong absolute number paper over a cost trend worth explaining on its own terms, not folding into the celebration.

Key Financial Metrics

Q2 2026 vs. Q2 2025, consolidated IFRS figures ("Accounting P&L"), reported in USD - Nu reports natively in US dollars, so no FX conversion is needed

Metric	Q2 2026	Q2 2025	YoY
Total revenue (IFRS)	$5,513.2M	$3,668.5M	✅ +50.3%
Interest income and gains, net of losses	$4,760.6M	$3,128.7M	✅ +52.2%
Gross profit	$2,346.5M	$1,548.0M	✅ +51.6% (margin 42.6% vs. 42.2%)
Total operating expenses (net of other income)	$1,105.6M	$667.6M	⚠️ +65.6%, outpacing revenue
Income before income taxes	$1,236.3M	$879.4M	✅ +40.6%
Income tax expense	$175.2M	$242.4M	✅ Effective rate fell to 14.2% from 27.6%
Net income for the period	$1,061.1M	$637.0M	✅ +66.6% (management's own FXN framing: +49%)

Balance sheet metric	Jun 2026	Dec 2025	Change
Total assets	$82,753.4M	$74,893.9M	✅ +10.5%
Total liabilities	$69,501.7M	$63,572.3M	⚠️ +9.3%
Total equity	$13,251.7M	$11,321.6M	✅ +17.0%
Deposits	$45,328.4M	$41,925.1M	✅ +8.1%
Credit card receivables + loans to customers (net)	$32,811.8M	$27,689.4M	✅ +18.5%
Cash and cash equivalents (quarter-end)	$13,551.6M	$15,003.6M	⚠️ -9.7% (deposit and loan growth outpacing cash build - normal for a growing lender)

Gross margin held essentially flat (42.6% vs. 42.2% a year earlier), which is itself notable given Expected Credit Loss^» expense grew 46.4% YoY (to $1,482.2 million) against 50.3% revenue growth - credit costs are no longer outrunning revenue the way they did in nu/2026-03's comparable quarter. The effective tax rate fell to 14.2% from 27.6% a year ago, a smaller and more mechanical swing than the one flagged in the prior post - this quarter's filing doesn't disclose a single dominant driver on the scale of last quarter's $319.8 million deferred-tax benefit, so the lower rate reads as a continuation of the credit-provisioning-driven deferred-tax pattern already established rather than a new one-off.

Key Operational Metrics

• Customers: 138.9 million (Jun 2026) vs. 122.7 million (Jun 2025) - ✅ +13% YoY FXN, ~4 million net adds this quarter. Brazil reached almost 118 million customers with monthly activity surpassing 86% for the first time; Mexico reached 15.8 million (16 million as of July 2026, per management); Colombia surpassed 5 million.
• Monthly activity rate: 83.5%, essentially flat sequentially (83.4% in Q1'26) and up from 83.2% a year ago.
• ARPAC^»: approximately $17.1, up 22% FXN YoY from ~$12.5 and up again sequentially from $15.9 in Q1'26 - the same multi-year monetization trend tracked since nu/2022-03, now roughly 2.6x that quarter's $6.7 figure.
• Efficiency Ratio^»: 19.5%, worse than 17.6% in Q1'26 though still better than 21.3% a year ago - management attributes the sequential deterioration to real estate and marketing spend shifting from Q1 into Q2, plus continued international-expansion investment.
• Total credit portfolio: $39.4 billion (credit cards $26 billion, unsecured lending $10.3 billion, secured lending $3.1 billion), +37% YoY FXN and +5% QoQ.
• Risk-Adjusted NIM^»: 12.4%, up 290 basis points sequentially from 9.5% in Q1'26 and up from 9.9% a year ago, as Cost of Credit declined 9% QoQ on the normal second-quarter improvement in early delinquencies.
• Asset quality: the leading-indicator 15-90 NPL^» ratio improved 16 bps to 4.8%, mostly on seasonality; the lagging 90+ NPL ratio rose 35 bps to 6.9%, which management attributes to the seasonal migration of first-quarter early delinquencies into later-stage buckets rather than a new deterioration.
• Geographic segment note: still a single reportable operating segment (the CEO, as Chief Operating Decision Maker, reviews the Group on a combined basis) - unchanged since nu/2022-03. This filing's interim disclosures don't break out a geographic revenue table for this specific period the way the FY2025 and Q1'26 filings did, so a Brazil-vs.-Mexico-vs.-Colombia revenue split isn't independently verifiable from the source documents for this quarter.

Beyond the Usual

The gap between "Managerial" and audited revenue widened again, in dollar terms

The non-IFRS "Managerial P&L" framework introduced in Q4 2025 and first flagged here in nu/2026-03 is still running: for Q2 2026, Nu's press release headlines Managerial Total Revenue of $5,875.7 million and Managerial Gross Profit of $2,441.1 million, versus the audited IFRS ("Accounting P&L") figures of $5,513.2 million revenue and $2,346.5 million gross profit filed the same day - a $362.5 million gap, up in dollar terms from Q1 2026's $347.5 million gap (though roughly stable as a share of revenue, at 6.6% vs. 7.0%). As before, net income is identical either way ($1,061.1 million under both frameworks) - this remains a presentation choice about where revenue-based taxes and reclassified float/fee income sit above the line, not a change to the bottom line - but a second straight quarter of a larger absolute-dollar gap means this needs to keep being tracked against the audited figures independently rather than assumed to be converging.

Legal-provision additions in Brazil nearly doubled year-over-year

The balance of provisions for civil, tax, and labor lawsuits grew to $44.3 million at quarter-end from $30.9 million at the start of the year, but the more telling number is the flow: new additions to the provision totaled $27.4 million in the first half of 2026, versus $14.5 million in the first half of 2025 - close to double. Civil lawsuits are described as "mainly related to NuAccount operations," and the company still classifies a further $4.8 million of civil claims and $4.9 million of labor claims as only "possible" (disclosed but not provisioned, per note 25c). None of this is large relative to a $13.3 billion equity base, but a near-doubling of the pace of new legal-provision additions in a single year, on a business still adding tens of millions of new accounts, is worth watching rather than dismissing as noise.

$38.2 billion of undrawn credit is sitting almost entirely off the headline numbers

Nu's own risk-management disclosures show $38,163,940 thousand ($38.2 billion) of "unused limits" on credit cards and lines as of June 30, 2026 - up from $28,841,327 thousand ($28.8 billion) at the start of the year, a 32.3% increase in six months. That figure isn't recorded on the balance sheet at all; it only appears in the financial-risk-management footnote because it factors into the Expected Credit Loss calculation. For scale: $38.2 billion in undrawn commitments is now nearly as large as the entire $39.4 billion on-balance-sheet credit portfolio the company headlines every quarter - a genuinely large pool of contingent future credit exposure that grows in near-lockstep with the visible book but gets none of the same attention.

Lease liabilities more than doubled as Nu builds a physical footprint for the first time

Lease liabilities rose to $66.4 million from $29.2 million since year-end (+127%), with the corresponding right-of-use assets more than doubling too ($54.7 million from $22.2 million). For a company that built its brand on having no branches, a lease footprint growing this fast in the same half-year that Nubank Mexico secured its full banking-operations authorization is a plausible, if unstated, connection - a licensed bank in Mexico needs more physical infrastructure (registered offices, back-office space, potentially limited-service locations) than a purely digital credit-card issuer did.

A $500 million buyback ran through half its authorization in six months

Nu repurchased 40,659,600 Class A ordinary shares during the first half of 2026 for a total of $500.4 million, leaving $499.6 million remaining under the program's authorization. This is a genuine capital-allocation signal worth sitting next to the record profit: a company that just posted its first $1 billion profit quarter and closed a Mexican banking license is also actively buying back stock rather than deploying that full pace of capital solely into new-market expansion.

Target Valuation Range

Nu closed Q2 2026 at $13.36, above the ~$11.36 bull-case fair value implied by a sustainable-P/B framework built on this quarter's ROE against its cost of equity - and further above the ~$7.01 base case and ~$3.98 bear case - even though its trailing-twelve-month EV/Revenue (~2.9x) and P/E (~17.9x) both compressed from the ~3.0x and ~20x annualized proxies used last quarter. Read as richly valued: the market is pricing in either a higher sustained ROE than this quarter's 33%, or a lower cost of equity, than a Gordon-growth framework can support on its own - with the Efficiency Ratio deterioration flagged above the likelier swing factor if that gap closes the wrong way.

Nu closed Q2 2026 at $13.36 (June 30, 2026) - down 24.7% from the $17.75 high hit just five months earlier (January 2026), and roughly flat versus a year ago ($13.72 at Q2 2025's close, -2.6%), a round trip already covered in more detail in nu/2026-03. With 4,830,688,659 total shares issued and outstanding (Class A and B combined, no stock split since the pre-IPO split already reflected in nu/2022-03's share count):

Market cap → enterprise value	Q2 2026
Share price (period-end)	$13.36
Shares outstanding	4,830,688,659
Market capitalization	~$64.5B
Total liabilities (interest-bearing borrowings, financing, leases)	~$4.7B
Less: cash and equivalents	~$13.6B
Enterprise value	~$55.7B

Peer-multiple sanity check	Q1 2026 (annualized proxy)	Q2 2026 (TTM, first real figure)	Change
Enterprise value	~$60.5B	~$55.7B	✅ down
TTM Revenue	n/a (proxy: $19.9B)	~$19.3B (Q3'25-Q2'26 actual)	-
EV/Revenue	~3.0x	~2.9x	✅ roughly flat/down
TTM Net Income	n/a (proxy: $3.5B)	~$3.6B (Q3'25-Q2'26 actual)	-
P/E	~20x	~17.9x	✅ down
P/B	~5.6x	~4.9x	✅ down
ROE (quarter, disclosed)	29%	33%	✅ up

Q2 2026 is the first quarter with a genuine trailing-twelve-month figure to compute against (Q3 2025 through Q2 2026, all four quarters verified from filed statements), replacing the annualized single-quarter proxy used in nu/2026-03 - and the real number is a touch better than the proxy suggested, not worse. TTM revenue of ~$19.3 billion and TTM net income of ~$3.6 billion put EV/Revenue at ~2.9x and P/E at ~17.9x, both cheaper than three months ago despite the business growing faster over that same window. P/B fell to ~4.9x from ~5.6x on the same equity-growth dynamic (equity up 17.0% since year-end while the share price barely moved).

A 33% ROE clearing an estimated 13-15% cost of equity by such a wide margin is real economic profit - the ROE-vs-cost-of-equity read is the honest lens for a bank, and a sustainable-P/B framework converts that into an actual implied price rather than a qualitative label. Book value per share was $2.74 (total equity of $13,251.7 million over 4,830,688,659 shares) at quarter-end; running that through a Gordon-growth justified-multiple (Justified P/B = (ROE - g) / (cost of equity - g)) across three scenarios:

Sustainable-P/B scenario	ROE (assumed sustainable)	Cost of equity	Terminal growth	Justified P/B	Implied price
Bear	20%	15%	4%	~1.5x	~$3.98
Base	28%	14%	5%	~2.6x	~$7.01
Bull	35%	13%	6%	~4.1x	~$11.36

Even the bull case - ROE climbing past this quarter's already-record 33%, paired with the most optimistic reasonable cost-of-equity/growth combination - implies a fair value of roughly $11.36, still below the $13.36 actual close. The stock's actual 4.9x P/B sits above the top end of every scenario tested here, which is a genuine tension with the compressing nominal multiples in the peer-multiple table above: the multiples got cheaper quarter over quarter, but on a fundamentals-derived basis the stock still trades rich to what a 33% ROE can support in perpetuity. That gap narrows only if the Efficiency Ratio deterioration flagged above reverses and ROE keeps climbing toward the bull case rather than mean-reverting toward the base or bear cases.

---