In case you didn't know this about me already, I'm an immigrant.

I was born in Cuba, lived in Russia until I was five, then immigrated to the US. I've spent my entire life navigating financial systems that weren't built for people like me: people who live between economies, send money across borders, and don't fit neatly into credit bureau risk models. So when I read my friend Andrew Dresner’s take on why Nubank will struggle in the US market, I see the analysis through a different lens.

I came across Andy's newsletter by chance. I quickly devoured the content and appreciated the depth and care he put into it. On a whim, I sent him a LinkedIn note saying I liked his newsletter, and asking if he'd be open to connecting. To my surprise, Andy said yes. Andy is a payments industry veteran and was previously a partner with McKinsey. Now we have standing meetings on Fridays to talk "newsletter" shop. All this to say - reach out. You never know what can happen.

He's looking at the market through traditional banking segmentation: expat populations, ethnic niches, product features. I'm looking at it through lived experience: 62 million US Latinos, $63 billion in annual remittances to Mexico alone, and decades of being underserved by institutions that fundamentally don't understand cross-border financial lives. Here's why I think the opportunity is bigger than the analysis suggests, backed by data, not just anecdote.

This platform will only get bigger with your support. Consider becoming a paid subscriber. You'll get exclusive access to emerging market dealflow, our full research archive, early previews of upcoming pieces, and the ability to shape future topics.

1️⃣ The remittance corridor is the wedge, not the market.

Andy's original analysis focuses on Brazilian expats (0.5-2.1M) as if they're the total addressable market. I think that's the wrong frame. The Brazil-US remittance corridor moved $2.8 billion in 2023, up from $2.1 billion in 2019 (World Bank). Mexico-US alone is $63 billion annually. These flows represent relationships, not residents.

Yes, the US-Mexico corridor is heavily served by Western Union, MoneyGram (through Walmart/Sifra), Remitly, and Wise. And yes, Nubank hasn't proven it will be cheaper. But here's what those services can't solve: the cash-out problem on the receiving end and the banking relationship on the sending end being fundamentally disconnected. My family still sends money back, and we use Western Union not because it's good, but because recipients need in-person cash pickup and we trust it won't fail.

That's the bar: "won't fail."

Not "delights me as a customer."

Nubank's opportunity isn't to be another remittance app. It's to be the primary banking relationship for the 45% of US Latinos who send remittances (Pew Research, 2023), where remittances are just one feature of a full financial relationship. The wedge is cross-border functionality. The market is everyone with a foot in both economies—62 million people who need checking, credit, savings, AND cross-border capabilities in one place.

2️⃣ “Already served” doesn’t mean well served.

The claim that US lenders serve every credit niche better than Brazilian lenders misses how expensive that service is. The average APR on US credit cards for subprime borrowers is 24.5% (Fed data, Q4 2024). BNPL and EWA have helped, but BNPL default rates hit 10.5% in 2024 (TransUnion), and merchants pay 3-6% per transaction. These aren't solutions. They're expensive patches on a broken credit infrastructure.

Nubank's Brazilian model isn't about serving the underbanked. It's about serving them profitably at scale. Their 28.3% efficiency ratio vs. Chime's ~100% isn't a temporary advantage. Andy's right that this comparison may not be fair. After all, Nubank will need to fund card loans through securitization, deposits, or warehouse lines. Most fintech lenders want deposits because they're cheaper and more stable (see SoFi). Wholesale funding is expensive and disappears when you need it most.

But here's the thing: Nubank already knows this. In Brazil, they've built a deposit base of over $20 billion while maintaining that efficiency ratio. Credit cards generate interchange and interest revenue from day one, creating immediate monetization that funds deposit acquisition. The model works: post a profit on credit, use that to attract deposits, use deposits to fund more lending at lower cost. They've proven the flywheel at scale. The question isn't whether it's possible, it's whether they can execute it in the US. That's a different kind of risk.

3️⃣ The credit-first model is actually superior in the US.

The question: can Nubank out-underwrite Capital One, arguably the most sophisticated underwriter on earth? Brazilian incumbent banks weren't credit experts like COF. And if Nubank can, how long before COF just copies their techniques?

I think Nubank has a structural advantage that's harder to copy than Andy assumes. Their ML-based underwriting was built from day one in a market with weaker credit infrastructure. They've underwritten 105 million customers with default rates under 6% in an environment where traditional FICO-based models don't work. Capital One is sophisticated, but it's sophisticated at underwriting within the US credit bureau system. Nubank is sophisticated at underwriting outside traditional credit data.

For US Latinos—many of whom have thin files, irregular income, cross-border employment, or family structures that don't fit W-2 models—that difference matters. The average American household has 3.5 credit cards vs. 1.5 checking accounts (American Bankers Association, 2024). Credit cards drive 78% of payment volume in the US vs. 34% in Europe (Nilson Report, 2024). Nubank starting with credit is better suited to US consumer behavior.

The challenge is execution, not product-market fit.

4️⃣ The language barriers are overstated; trust barriers are understated.

The Portuguese vs. Spanish distinction matters less than the analysis suggests. I grew up speaking Russian and Spanish in a Cuban household. Language wasn't the barrier to financial services. Trust was. When you're an immigrant, you don't trust banks that treat you like a risk score. You trust institutions that understand your life looks different: money moving between countries, inconsistent income documentation, family members without SSNs.

Andy's point is fair: there's no proof this is how Latino consumers will behave. The affinity is asserted, not proven. And it can't be proven until they try.

But here's what we do know about why the Latino population is underbanked. Part of it is income. Part of it is documentation—many aren't documented, and while major banks accept Consular Matricular cards for deposit accounts, KYC remains a hurdle. But part of it is also cultural understanding of money and institutions. In Russia, there's a word for money you put under your bed: "it's for the black day." When your lived experience is economic volatility and institutional failure, trust matters more than features.

Nubank's actual moat isn't language. It's that it was built for this. Brazil has a 140 million unbanked/underbanked population. Nubank figured out how to serve them at a 28% efficiency ratio while posting a profit. The behavioral proof will come from execution, but the structural advantage is real.

5️⃣ The US Latino market isn’t saturated. It’s underserved at profitable unit economics.

Yes, there are Latino-focused digital banks. Comun raised $24M. Openbank operates in the US. Banco Santander has initiatives. None of them have Nubank's economics. Comun is pre-revenue and focused on remittances (classic wedge, no monetization engine). Openbank is a savings account play (low revenue, high customer acquisition cost). Santander USA has a 60%+ efficiency ratio and can't cross-sell digitally to save its life.

The US Latino population is 62 million and growing at 2.8% annually (Census Bureau). Latino households hold $2.8 trillion in purchasing power (2024), up 87% from 2010. Yet they remain systematically underbanked: 60% are underbanked vs. 20% of white households (FDIC, 2023). That gap represents 37 million people. If Nubank captures 5% of that at their current LTV/CAC ratio, it's a $50B+ market cap business in the US alone.

6️⃣ Incumbent structural limitations create advantage - but marketing costs remain the unknown.

All major banks have online and mobile channels that can open and service accounts. The efficiency ratio for the marginal cost to serve a fully online customer at one of those banks is actually lower than Nubank could hope to achieve due to scale.

And branches exist because consumers want them. Fifth Third is building over 100 branches per year to meet that demand. In the absence of branches, digital banks need paid marketing. That's what gives Chime an awful efficiency ratio, and the same applies to every digital bank. Even USAA—with one of the strongest affinities—is fully digital but has low efficiency ratios and high CAC.

The marketing cost question is real. I'm asserting that Latinos will flock to a fully digital affinity bank, but Nubank will have to spend marketing dollars to make that happen. There's no viral credit card service that doesn't spend a fortune on marketing.

But here's where I think the structural advantage still holds: Nubank doesn't need to compete on marketing spend dollar-for-dollar. They need to achieve better conversion from marketing spend because the product solves a real problem incumbents don't. My family uses Western Union not because we saw a good ad, but because nothing better exists that we trust. When something better exists, word of mouth in immigrant communities is real. We talk to each other.

US banks can't replicate Nubank's model because they're structurally incapable in other ways:

Nubank has none of this. It was born digital, in a market that forced extreme capital efficiency, with ML-based underwriting from day one. The marketing cost is the wild card. But the product-market fit advantage is structural.

7️⃣ The real comparison isn’t Chime. It’s Synchrony and Capital One.

The analysis compares Nubank to Chime, but Nubank isn't a neobank. It's a credit-led digital bank. The right comparison is Synchrony (credit card specialist, $82B in loans, 46% efficiency ratio) or Capital One's credit card business (33% efficiency ratio).

Andy asks: if Nubank has to compete with COF, why can it do so for less? The answer has to be that strong affinity lowers marketing costs. And he's right—that's asserted rather than proven. It can't be proven until they try.

But Nubank is posting better economics than both Synchrony and Capital One—28% efficiency ratio vs. their 33-46%—with faster growth (23% YoY customer growth in Brazil, Q2 2025), in a less developed market, while building deposits and investment products on top of the credit base. If they port even half of that execution to the US Latino market, they're not competing with Chime for neobank scraps. They're competing with Capital One for the future of consumer credit in an underserved segment.

The immigrant economy is not a niche.

Here's what traditional banking analysis misses: immigrant households don't think in terms of "ethnic banking." We think in terms of who makes our lives easier across two economies. My family doesn't use "Russian-American banks." We use whatever works, which is usually a patchwork of Western Union, wire transfers, and Telegram coordination because nothing is built for us.

Yes, we use Western Union despite it being mediocre. Why? Because recipients need in-person cash pickup and we need it not to fail. That's not rational optimization. That's risk mitigation born from lived experience where every unfamiliar thing has failed before. The barrier to adoption isn't product features. It's trust built over time.

Nubank has a 10-year head start building that product. They've done it at scale (105 million customers). They've done it profitably (28% efficiency ratio, 15% ROE). And they've done it in markets that look more like the US Latino population than the US looks like Chase's customer base.

The question isn't whether the Brazilian expat market is big enough. It's whether Nubank can execute a credit-first, digital-native, cross-border banking model in a market with 62 million Latinos, $2.8 trillion in purchasing power, and zero incumbent capable of replicating their unit economics. They haven't proven the marketing costs will work. They haven't proven the affinity will translate. They haven't proven they can out-underwrite Capital One.

But I think they can. And I think "ethnic bank" dramatically undersells what they're building.

🎧 Last week I re-aired my conversation with Ehis Akhetuamhen about his powerful journey to the US—a story that speaks to the immigrant experience in ways data never could. Coming up on February 11th, my conversation with Jeff Hurst from Furnished Finder will air, exploring how he built a business serving a market others overlooked.
👉 Listen on NPR, Apple, Spotify, or wherever you listen to podcasts

Ask: I am looking for a place to record in LA for upcoming episodes. Ideally something with good acoustics and privacy, whether it is a studio, podcast space, or rentable office. If you have any leads, please let me know.

📌The wild markets behind Polymarket’s truth machine

📌Apple Music kicks off Bad Bunny’s Road to Halftime ahead of Super Bowl LX

If you’ve made it this far, I hope I’ve convinced you to become a paid subscriber!

Till next week,
Ilona

“When I first opened Attio, I instantly got the feeling this was the next generation of CRM.”
— Margaret Shen, Head of GTM at Modal

Attio is the AI-native CRM for modern teams. With automatic enrichment, call intelligence, AI agents, flexible workflows and more, Attio works for any business and only takes minutes to set up.

Join industry leaders like Granola, Taskrabbit, Flatfile and more.

👉 Start for free today