History
The Narrative Arc
Nu's story has bent without breaking. From IPO-day skepticism in late 2021 — when underwriters cut the price range from $10–$11 to $8–$9 days before pricing — the company has compounded customers from 54M to 131M, swung from a $364M 2022 loss to $2.9B of 2025 net income, and never changed founders. The founder-CEO who promised "the world's largest digital banking platform outside Asia" in 2021 is now selling a "rebuild banking around AI" thesis and a U.S. bank charter. Management has earned the right to be heard, but the next chapter — global expansion and AI transformation — is the first where the bear case lives in the future, not the past.
Founder still in the chair. David Vélez has been CEO since founding the company in 2013, and controls 88.3% of Class B shares (74.4% of voting power). Every credibility test in this section is about his track record — there is no "new management inherited a good business" story to tell here.
The chapter that defines the current franchise began in 2023 — the year the loss-making challenger became a structurally profitable retail bank. Everything before that is the asset; everything after is the test of whether the team can deploy it.
What Management Emphasized — and Then Stopped Emphasizing
Four discrete eras show up clearly in how management talks about the business. The phrase that anchors each year-end letter has rotated almost on schedule, and what was loud in 2022 ("path to profitability") is now almost silent, replaced by "AI transformation" and "global digital banking platform."
Management emphasis on each topic, scored 0 (silent) to 5 (anchor theme).
Three patterns matter:
Loud, then silent. "Path to profitability" disappeared the moment profitability arrived. So did "crypto" — promoted heavily in 2022 when NuCrypto launched, now a back-page mention. Insurance and NuInvest get one paragraph in the 2025 letter where they got pages in 2021. Management isn't hiding these — they simply stopped being the story.
The quiet swap. Through 2023, the headline number was customers. From 2024 forward, the headline is ARPAC per customer ($11.2 → $15.9 in seven quarters). This is the right pivot — saturated markets need monetization, not more signups — but it's worth noticing that it coincided with Brazilian net adds slowing from 25% to 14% YoY. The new metric flatters the new reality.
The two big additions. "AI transformation" went from a buzzword in 2024 to the anchor of the 2026 narrative, with nuFormer disclosed at parameter-count scale comparable to Meta's LLaMA. "US expansion" went from non-existent to the closing-remarks centerpiece of every 2025–26 call. These are real strategy shifts, not just framing.
Risk Evolution
The 20-F risk factor section is remarkably stable on the macro and regulatory front — Brazilian rate volatility, government intervention, and cybersecurity have been top-five risks every year. What has changed is what management worried about most: the existential "we may never be profitable" language has disappeared, and three new risks have surfaced.
Risk-factor emphasis in 20-F filings, scored 0 (absent) to 5 (top-tier).
What got smaller. The 2021 prospectus carried explicit "we have incurred losses since our inception, and we may not achieve profitability" language in its risk-factor summary. By 2024 that bullet point is gone from the summary section. The risk was retired by execution.
What got bigger. In 2024 and 2025, management added a brand-name risk — Brazilian regulators were consulting on whether non-banks could keep using the word "bank" in their name. That's a regulatory threat to "Nubank" itself, and it didn't exist in earlier filings. The 2025 20-F adds two new top-tier disclosures: AI operational/reputational risk (the first time it appears at all) and explicit Brazilian election risk tied to October 2026.
What stayed loud. Brazilian rate and FX volatility — the single most-cited macro risk — has not faded. Founder voting control has been a Class B risk every year since IPO, and Vélez's stake has shifted only marginally (88.6% → 88.3% of Class B).
How They Handled Bad News
Three episodes are worth comparing because the style of the disclosure tells you something about credibility.
The introduction of "Managerial P&L" in Q4 2025 is the first disclosure choice in years that deserves skepticism. Management now presents a non-IFRS managerial framework as the primary headline, with IFRS reconciliation in an appendix. Stated rationale is multi-segment comparability. Investors should watch whether managerial revenue/efficiency metrics diverge from IFRS in ways that flatter the trajectory.
Guidance Track Record
Nu does not give point-form earnings guidance, which makes "promises" harder to grade. But management has put specific, valuation-relevant claims into the public record over the years. Here is the scorecard on the ones that mattered.
Credibility score (1-10)
Credibility score: 8 / 10. Every binary promise that's already had time to mature — profitability, ROE, Mexico break-even, efficiency-ratio compression — has been delivered, often ahead of internal plans (Q1 2026 IFRS-profit milestone "arrived ahead of our own internal plan," per Vélez). The two big TBDs — the "less than 100bps efficiency drag" U.S. cap, and the asset-quality narrative under AI-accelerated underwriting — are both future tests, not past misses. The CSA termination is the single strongest signal of CEO alignment in the file. The point off is for the Q1 2026 EPS miss, the introduction of a non-IFRS managerial framework as the primary headline, and the fact that the hardest promises (US, AI economics) haven't been tested yet.
What the Story Is Now
The 2026 story is in three parts: a profitable Latin American core, an asymmetric U.S. option, and an AI-transformation bet on whether banking gets re-platformed.
Believe. The Latin American franchise is the rare combination of high growth, structural cost advantage, and proven credit underwriting through a 110-bp NPL cycle. The team that built it is the team that still runs it. Promises about profitability, ROE, efficiency, and Mexico break-even have all been delivered, on or ahead of schedule.
Be skeptical of. The "AI-first bank" framing and the U.S. expansion claim of bounded downside. Both are forward-looking, both are unfalsifiable in the short term, and both arrived in the disclosure just as the "path to profitability" story retired. That timing — replacing one narrative pillar with two new ones precisely when the old one stopped working as a forward driver — is the pattern to watch.
The story is simpler than it was in 2022 (when the company was a customer-count chart with no earnings). It is more stretched than it was in 2024 (when Brazil execution was the whole thesis). Where it goes from here depends less on whether the LatAm flywheel keeps spinning — it will — and more on whether the U.S. and AI bets compound the franchise or merely consume capital while the core matures.