Nu Holdings Ltd. (NU) Earnings Call Transcript & Summary

Thanks, operator, and I appreciate all of you being here on today's earnings call. Joining me are David Velez, our Founder, Chief Executive Officer and Chairman, as well as Rob Livingston, our Chief Financial Officer. Every financial metric we discuss and present today is based on our mini geo P&L framework, introduced back in the Q4 2025. These managerial measures matter greatly to how we run the business, yet they don't qualify as financial measures under IFRS and might not line up with those of other companies. You can find a complete reconciliation report to the most directly comparable IFRS figures in our managerial P&L reconciliation report and in the appendix to this presentation. Except where stated otherwise, every growth rate covered today is shown on a year-over-year FX neutral basis. This discussion could contain forward-looking statements, which aren't guarantees of future performance and carry risks and uncertainties. Actual results might differ materially from those expressed or implied. For further details, please see the forward-looking statements disclosure in the earnings presentation; with that, I'll hand things over to David. Go ahead, David.

Good morning, everyone, and thanks for being here with us today. Thirteen years back, we began with one straightforward idea. A financial institution powered by technology, without physical locations or an entrenched legacy business to protect, could reach hundreds of millions of individuals more effectively and at a much lower expense. Now, I am pleased to share that during the most recent quarter, for the very first time, we produced over $1 billion in net income; this achievement stems from our dedication to customers, which has turned into a formula that generates earnings. It also reflects the incredible efforts of our colleagues at Nubank. Thirteen years on, that original idea keeps unfolding precisely the way we imagined it would. Our customer base has reached 139 million, with Brazil accounting for nearly $118 million of that and Colombia contributing more than $5 million. Mexico, meanwhile, had just crossed the 16 million customer mark by the end of July. Engagement grew stronger right alongside this expansion. The activity rate climbed quarter over quarter, landing at 83.5%, and Brazil passed 86% for the very first time. Monetization keeps getting powered by this combination of a larger customer base and deeper engagement, with RPAC reaching $17. Together, those elements generated $5.9 billion in gross revenue, and throughout it all we maintained a highly efficient operating model, as shown by an efficiency ratio of 20%. Thanks to this operating leverage, we can continue directing funds into our 3 core markets. Brazil, Mexico and Colombia, as we also build the groundwork for growing internationally; that is precisely why we have always focused on optimizing for the long run. It explains how we can keep building for the coming decade while also delivering a quarter such as this one. Allow me to take you through both, beginning with Brazil. Brazil is still our biggest growth opportunity, and the bulk of it sits inside our current customer base. The mass market by itself accounts for about $30 billion in industry gross profit; we already cover most of the segments, and we serve as the primary account for roughly 60% of those customers. Even with that, there is plenty of space to strengthen those ties and capture additional profitability; this is achievable thanks to the capabilities we have developed across the last 13 years. Those capabilities enable us to broaden financial access while providing an improved customer experience, reduced costs and products that are increasingly tailored. As we established one of the top financial services brands in Latin America aimed at the mass market, we discovered that we were drawing in millions of higher income individuals that, regrettably, we could not serve properly back then; in 2021, we introduced Ultravioleta, a brand and product aimed at higher-income customers, and nearly 1 million Ultravioleta customers have notably larger purchase volumes and assets under custody compared with the remainder of our portfolio, with both still expanding robustly, rising 41% and 37% year-over-year, respectively, in Q2 2026. Yet we came to understand that a substantial segment exists between the mass market and high income that we could be serving better too. We refer to this segment as Supercore; and in July, we rolled out Chroma, a subscription-based tier designed for our Supercore customers. There's another segment where the profit pool is actually bigger than high income, and it's one where our penetration is already substantial. With Chroma, they get a tailored experience, better credit options, and a wider range of banking and lifestyle perks — all built to incentivize customers to bring more of their financial lives over to Nubank. That covers things like a new cell plan, a ChatGPT subscription at no cost, savings products with accelerated returns, and additional perks through partners in our own ecosystem. What we're aiming for is to build primary banking relationships, and for this segment, Chroma represents a major step toward that goal. Naturally, the opportunity isn't limited to consumers either. We currently serve 6.8 million small businesses, which makes Nubank the biggest financial institution in Brazil measured by business customer count — even though we've only reached roughly 1/3 of that market. That's the way we look at the extraordinary growth available in Brazil: keep growing our customer base while serving a bigger portion of their financial lives through improved products and segmentation. Let me now shift to our other core markets. This month, Mexican regulators gave us the green light for our banking license, and we're excited to start operating as Mexico's biggest digital bank, serving more than 16 million customers. With this, our transformation is finished — we've gone from a Fintech focused on credit first to a complete digital bank — and it unlocks abilities we simply didn't have before. When paychecks land directly into accounts, it strengthens the core banking ties people have with us and keeps them more engaged. Having more deposit insurance makes people feel safer about leaving their money with us; those funds let us extend more credit, and over time they allow us to branch into fresh products and new kinds of customers. Bringing people into the financial system has run through this whole story. We opened the very first bank account for 35% of the people we serve, and the first credit card for 52% of them. Today you'll find our customers in 98% of Mexico's municipalities, and close to 80% of them live outside the big cities — proof that technology lets us reach people no matter where they are. Even so, it's the future that gets us most excited. When it comes to adopting digital finance, Mexico is still fairly early in the game. Over the last ten years, bank account penetration has gone from 44% up to 63%. Only 85% of Mexicans still said they'd rather pay with cash. All the same, things are moving faster now, and digital payments in Mexico just keep growing year after year. In the first six months of this year, transfers under $5 increased by over 60%. Right now, close to 50% of all transfers made in this country are below $25. Such transactions happen daily, and they clearly indicate that cash is gradually being replaced by these payments. New regulations were rolled out by the Central Bank in June, and every financial institution is required to put them into effect before the year ends. The goal here is to make the experience simpler across various payment rails. From now on, customers will encounter a uniform interface and go through identical steps no matter who the recipient is or which payment method they pick. Because these rules apply across the whole financial system, they make network effects stronger and ought to speed up the shift toward digital payments even more. We have seen this exact pattern before. A regulatory nudge in Brazil sparked competition and digital innovation, widening financial access, increasing everyday use, and in time accelerating credit uptake. Pix is the clearest case: right out of the gate we offered a simple, seamless experience, climbed to the number one spot by transaction volume, and turned that into primary banking relationships; that situation fit hand in glove with the digital strategy we have built. In our judgment, Mexico is following a similar path. The numbers we have so far make that clear. As of now, we have gotten to 16.5% of Mexico's out of population, basically matching the penetration level Brazil showed in 2020. Yet these cohorts are generating revenue sooner. At the equivalent point, ARPAC stands at $12.3 for Mexico versus $5.6 for Brazil, thanks to greater income per capita, stronger unit economics on the credit card offering, and larger interest-bearing balances, while serving costs remain lower. Mexico represents Brazil's strategy executed more quickly and aided by the scale available to us now. That explains why we achieved break-even in 6 years in Mexico, whereas Brazil took 8 years. In summary, consumer habits, technology, and regulation are all now pointed the same way. Combined, they produce one of the most attractive opportunities we have ever encountered in Mexico.

With additional financial activity shifting to our platform, our ties with customers grow stronger, our underwriting insights improve, and we can serve a bigger portion of their financial lives. In Mexico, we now possess, for the very first time, the complete range of capabilities needed to seize that opportunity. Plus, Brazil and Mexico operate on one technology stack and, more and more, under one brand. Let me demonstrate what that means. Roughly 12 months ago, we launched our flotation model for financial behavior. Since that point, one goal has driven us: creating a single AI platform that drives business and customer decisions throughout Nubank. That effort touches each layer of the stack. We grew and enhanced our in-house GPU fleet, which gives us total control over the compute layer. We broadened our architecture research work, and we keep building on one of our biggest strengths: over 10 years of transaction history spanning upwards of 100 million customers in 3 countries. That research is producing compounding gains in efficiency and model quality. We recently moved the former to a hybrid linear attention — the same architectural strategy underlying front models such as KemiK3 and Quon3.5 — and we trained using Moen, the same optimizer class that powers today's most efficient large language models. By decoupling the former core backbone for particular downstream decisions, any enhancement to the central model can immediately boost performance across every business line we have, without retraining costs. Our newest generation quadrupled context length, training speed and inference speed, all while lowering the expense of running models in production. As we've scaled pretraining of the base models, their grasp of customer behavior has gotten deep enough to reshape how we build every model upon them. One example: these days, we get the same predictive performance from 20 million fine-tuning data rows that used to take over 400 million, shrinking development cycles from weeks down to days. Nearly every decision we make now flows through this platform. We initially rolled out nuFormer in our flagship credit portfolio in Brazil. Across 2025, we reproduced the model in Mexico, proving the platform generalizes between markets. In the first half of this year, we brought it to unsecured lending in Brazil and to the next generation of our core credit models. Right now we're piloting it in credit cards for SMEs and for customers in Colombia; yet underwriting is just one use case. These days, AI agents manage over 60% of customer support conversations in Brazil, with customer ratings equal to or better than human parity. Past underwriting and customer support, we apply artificial intelligence to fine-tune decisions spanning credit, deposits and growth, shifting from predictive outcomes to identifying the actions that maximize value given real-world constraints — and the same grasp of transactions that forecasts credit risk also forecasts what a customer will want next. This lets us suggest products that maximize long-term customer value, tailor the app experience, and advance toward our vision of a private banker. nuFormer is also enhancing the way we expand. Since the model picks up on our depiction of each customer's behavior, we rely on it to position every campaign before those customers who are most apt to consider it valuable, and over 100 campaigns have already been executed in this manner. A single platform drives underwriting, customer support, optimization and growth. each enhancement we introduce aids every application constructed atop it. We're extremely thrilled about the advancement Nubank has achieved so far, utilizing AI as its foundational technology and hold firm assurance that our strategy will deliver meaningful differentiation down the road. Prior to moving on to our financial results, I'd like to share a few remarks regarding our CFO transition. As we stated in early June, Rob Livingston has taken over from Guilherme Lago as our Chief Financial Officer. Lago was with us for 7 years, 5 of those as CFO, and he passed the baton during the most powerful point in our history with our first $1 billion quarter. He has been a phenomenal collaborator, and I'm happy we'll continue working together in his fresh capacity as special adviser. Rob has been spending the last few weeks collaborating with Lago and our teams, and we're quite eager to be able to respond with his considerable expertise. Rob, welcome. The floor is yours.

Thanks, David. It's an honor to take on this position at a pivotal time for the business. Since I came aboard at Nubank, I've devoted time to meeting colleagues throughout the organization. The thing that has struck me most is the devotion to customers, the steadiness of the business model, and the rigor with which it has been carried out. I'm eager to assist in steering the coming chapter of Nubank's story. And I'm glad today to take you through the financial results for Q2 2026. Let's begin with the consolidated credit portfolio. It hit $39.4 billion, which is 37% higher year-over-year and 5% higher sequentially. The growth stayed broad-based. Credit cards rose 35% year-over-year, reaching $26 billion. Unsecured lending climbed 45%, hitting $10.3 billion, while secured lending was up 30%, reaching $3.1 billion. Sequential growth stayed solid while settling down following a stretch of unusually powerful expansion; origination doesn't move in a straight line, and we view that as a good dynamic. All through the quarter, our approach to underwriting stayed the same, and growth stayed robust compared with the wider market. As we'll get into over the next few slides, we feel good about the portfolio's quality and how the vintages we're originating are performing. Now shifting to deposits; we finished the quarter with $45.3 billion in deposits, which is 18% higher year-over-year and 6% higher sequentially, bringing back the seasonal outflows we talked about last quarter. Brazil came in at $36.4 billion, Mexico at $5.7 billion, and Colombia at $3.3 billion. In Mexico, deposits slipped modestly once more this quarter, which mirrors our continuing deposit optimization strategy; this keeps enhancing our cost of funding while preserving ample liquidity, with our loan-to-deposit ratio still sitting at only 35%. Our cost of deposits was 88% of the interbank rate, basically flat versus last quarter and 3 percentage points below where it was a year ago. All in all, we're happy with both the growth and the pricing of our deposit franchise across all 3 markets. As ever, our goal isn't merely to maximize deposits but to create a resilient funding base that deepens customer relationships, supports profitable growth, and bolsters the long-term economics of the business. Net interest income came to $3.7 billion, up 9%, and net interest margin widened 180 basis points to 22.9%. That stems from what we set out last quarter: the growth we added to the books, a mix weighted more heavily toward unsecured lending, and the deliberate risk expansions we pursued. Cost of credit fell quarter-over-quarter to $1.7 billion. Desenrola, the government debt renegotiation program, affected this figure by only about 5%. And even more significant than the accounting effect, the program let us help nearly 1.8 million customers renegotiate past due balances and get their finances back on track. With margin rising and cost of credit falling, risk-adjusted net interest margin widened to a record 12.4%, up from 9.5%. I'll take you through the drivers behind this expansion in risk-adjusted net interest margin in greater detail. So now we move to the bridge for risk-adjusted net interest margin, and I'd like to zero in on how it widened from Q1 to Q2. Credit income served as the main engine behind that risk-adjusted NIM growth, adding 178 basis points to the sequential rise, versus 152 basis points during Q1. What drove that step-up was the robust expansion in our card and unsecured lending books during Q1, which shows up in the better loan-to-deposit ratio we're seeing. A reduced cost of credit gave us another 115 basis points. Most of that cost of credit gain stemmed from the seasonal patterns we'd anticipated along with consistent performance in the underlying business, not from onetime items. Float income and funding costs each stayed essentially flat. Put together, these forces account for the significant risk-adjusted NIM widening this quarter and keep showing how resilient our underlying unit economics are, which backs the durability of present levels as we move ahead. Let me now shift to asset quality. Our NPL metrics, as anticipated, are still tracking their usual seasonal shape. Delinquencies in the 15- to 90-day bucket got better by 16 basis points, landing at 4.8%. That gain mirrors a number of dynamics underneath, and I'll break those down on the following slide. Delinquencies of 90-plus days rose 35 basis points to reach 6.9%, largely mirroring the seasonal shift of early delinquencies from the first quarter into the 90-plus category. Considered as a whole, these figures line up with the seasonal behavior we'd projected and keep reinforcing our belief that the portfolio's underlying quality is still solid. Turning to what's behind early delinquency, this bridge shows why the 15- to 90-day ratio got better on a sequential basis. Seasonality pulled the ratio down by 37 basis points. Offsetting that, the deliberate risk expansions we made during the first half of the year pushed it back up by 24 basis points. Those were conscious choices to serve groups with greater expected losses, but which likewise produce stronger risk-adjusted returns, as the performance of our risk-adjusted margin has shown. Product mix and the other drivers are roughly flat; we find no sign anywhere in our portfolio of a widespread softening in consumer credit, though we stay watchful as we always do. On balance, the 15- to 90-day ratio got better by 16 basis points over the quarter. From the balance sheet angle, the allowance bridge tells much the same tale. The allowance climbed from $6.1 billion up to $6.6 billion. Portfolio growth was by far the biggest factor, adding $342 million. Because of IFRS 9, we book expected credit losses when loans are originated. That means growth pushes the allowance higher before the matching interest income comes in. The deliberate risk expansions we just went over added a further $170 million, while every other change was immaterial, Desenrola included, since it mainly hits recoveries instead of the ECL allowance because of how renegotiated loans are accounted for. Taken together, these 2 bridges drive home the same point. This quarter's credit behavior came from growth, seasonality and disciplined risk expansion, not from any erosion in the portfolio's underlying quality. Our stance on provisioning and coverage stays disciplined and consistent. Now, looking at the chart on the left side, our allowance build came to 113% of newly formed 15-plus delinquencies for the quarter, which is roughly consistent with what we've seen historically. The right-hand chart makes the very same point. Our total coverage ratio over NPL90+ was 244%, so we're still holding allowances that equal nearly 2.5x our 90-plus balances; that gives us a solid balance sheet buffer and stays in line with what we've kept for the last several years, even while the portfolio keeps expanding. Those 2 metrics together drive home something important: as the portfolio grows across products and customer segments, our approach to provisioning stays disciplined and consistent throughout the cycle. Now, one last item on credit risk, and it matters. As David said, Nubank holds the leading position in Brazil when it comes to primary banking relationships. That position speaks to the trust customers have in us. Paired with the analytical strength of our underwriting models and the data quality those relationships produce, it gives us a structural advantage in credit. This slide makes it clear — it shows the share of credit card outstandings in Brazil that are 90-plus days delinquent. Our credit risk results have held steady across each income band. Behind this strong, stable performance are customers who count Nubank as their primary banking relationship. For those customers, delinquency rates run at about half of the portfolio average.

What's the takeaway here? Being the primary bank for customers doesn't only fuel expansion and top-line results — it also gives us an edge in credit. Deeper ties yield more detailed behavioral information, strengthening our underwriting and securing Nubank's position at the top of the payment hierarchy for those we serve. Combined, these elements produce reliably superior credit results. That's why customer primacy remains a cornerstone of our credit superpower, in our view. Let's turn now to the income statement. We recorded gross revenues of almost $5.9 billion, a 39% increase compared to the same period last year. Gross profit reached $2.4 billion for the quarter, up 43% year-over-year and 25% quarter-over-quarter. Since credit followed its usual seasonal trend, its gross profit contribution rose 41% in the quarter, while all fees made up 25% and flow accounted for 34%. Those proportions shift around quarter by quarter, as you'd expect. The key point is that each of the 3 components continues to expand in absolute dollars, showing how diversified our business model is. Looking ahead, we're confident that credit, fees and float will continue working together to drive long-term profit growth. Moving to operating leverage: net revenues passed $4 billion for the first time, reaching $4.1 billion, an 8% increase from the prior quarter. On the expense side, we landed at $806 million, which represents a 20% increase versus the prior quarter. The sequential rise reflects real estate and marketing costs shifting out of Q1 and into Q2, plus our ongoing spending on international growth. That pushed our efficiency ratio to 19.5% for the quarter. As we mentioned three months ago, the 17.6% figure we posted in Q1 wasn't something to model forward. Roughly two-thirds of the gain there stemmed from timing-related items that were always going to unwind, and that's exactly what happened this quarter. Going forward, we continue to anticipate a full-year efficiency ratio averaging roughly 20%. And more significantly, nothing has shifted in how we think about the long term — we'll keep funding growth while preserving operating leverage as the business gets bigger. So, to close things out: for the first time ever at Nubank, net income reached $1.1 billion. That's a 17% increase over the first quarter, and 49% higher than the same period last year. What matters even more is that we achieved this while keeping return on equity at a record 33%, and while still putting money into all 3 of our markets as well as our long-term opportunities. In my view, this quarter demonstrates the strength of the business model that David walked through earlier. The money we've put into technology, AI, underwriting and customer experience keeps turning into profitable growth at scale. Being part of this next chapter is thrilling for me, and I can't wait to keep building on this momentum. With that said, let's open the call for your questions.

Thanks, operator. Would you mind bringing Mr. Tito of Goldman Sachs onto the line.

I get the sense there are several different pieces at play here, Rob, and you threw out a few figures. My recollection is that you said something like 5% of provisions, though I'm not certain whether that actually flowed through to the bottom line. And I'm just reflecting on the fact that during our earlier calls, you'd noted risk-adjusted margin recovering to 10.8%, and this quarter it's well above that — clearly a very strong result. Still, just so I understand, did this enroll out have any effect on that risk-adjusted margin, given the various components involved? Then I have one more question after this.

Sure. I really appreciate that, Tito. I'll handle that one, and it's nice to hear from you once more. Correct — I did say this enroller affected our cost of credit by roughly 5%. That's the key measure we focus on there. Consequently, it likewise affected our risk-adjusted net interest margin. When you consider the improvement we delivered between last quarter and this one — nearly 3% of expansion in that metric — most of it was driven by growth in lending, and that largely stemmed from the robust growth we experienced in Q4 of last year and Q1 of last year, plus the way that matriculated into revenues during Q1. On top of that, the cost of credit added 115 basis points. And you're correct that this enroller would be a piece of that; it accounts for a minority of the effect. Seasonality was the biggest contributor, along with genuinely strong underlying credit performance.

All right. No, that's quite useful, Rob. And I suppose, to frame it in context, the thing weighing on everyone's mind is what happens as we head into next year. You said that on the whole, credit quality trends are looking favorable and you're at ease. Yet with the macro backdrop we're witnessing in — how do you view the growth picture, perhaps as we move toward 2027, along with this degree of risk-adjusted net?

David speaking. As we've told investors on several occasions, we don't necessarily adopt a directional stance regarding the economy. When we underwrite a loan, our foundational premise is that tomorrow will be worse than yesterday. Circumstances will in fact be far worse than anything we've witnessed; so a great deal of — each and every underwriting choice already presumes deterioration as the default. There's a fairly substantial cushion regarding what would need to occur for that choice to remain NPV positive. Clearly, we're operating in a setting marked by considerable caution; we're closely monitoring every signal available to us; thus far, we genuinely don't detect any meaningful or structural erosion in our figures, and we keep functioning with a very considerable cushion in this setting; we're additionally situated such that we — although our consumer base is large, we remain a very tiny fraction of the market. As we noted on the previous call and on this one, we hold 7% market share of that profit pool. So we remain a minor participant in that sizable market, and we're able to cherry-pick our customers. Cherry-picking them with loans and products of very short-term duration, which grants us an enormous capacity to respond swiftly and provides plenty of conviction regarding many of the underwriting capabilities, as we discussed. And then chiefly, we possess this enormous edge, as Rob noted, of being the biggest primary bank accounting in the country at present. More than 60% of our mass market customers rely on us for their primary bank accounts. So that's an enormous edge because, as Rob noted, it positions us effectively as senior in a customer's credit stack. And so once you combine robust analytics, a considerable cushion, being the primary bank account, and holding that seniority, then plenty of conditions exist for us to keep growing at a very solid pace. I won't give you a particular growth figure; we continue to see the conditions for continuing to grow and take share as we deploy many of these levers to do so very effectively.

Yeah. As for what you asked about net interest margin, or risk adjusted net interest margin, looking ahead as far as we can see, we expect it to stay around the level it's at right now. Our view is that this can hold.

Operator, could you please open the line for Mr. Jorge Kuri from Morgan Stanley.

Hello, everybody. I appreciate the chance to pose some questions, and congratulations on those impressive results. I want to revisit, I suppose, that risk-adjusted NIM figure of 12.4%. Perhaps I can connect that to how you're deploying AI, the growing sophistication behind your credit line increases, and generally your capacity to accept greater risk while experiencing smaller losses. Given that you've now observed what may be a full year's vintage of customers whose offers were enhanced through AI models, and you've kept tabs on them, could you walk us through a few of the KPIs you've noticed? Also, how closely do those relate to the 12% risk-adjusted margins, since you clearly sound quite assured that this represents the new normal? You mentioned it just now, Rob, and you also did so during an earlier Bloomberg interview. Historically, we've witnessed that figure fluctuate because of mix. So, yes, I'm simply hoping to hear a bit more about which KPIs are behind it. How much do the AI models contribute, and how much reassurance can we take that this level acts as a floor going forward?

Yes. Many thanks for that question, Jorge. There are a few points I'd like to make. To begin with, it's undeniably true that our generative and assisted models outperform conventional logistic regression models. Even so, we monitor them in precisely the same manner we would have monitored the models we used previously. We examine how predictable they are, the variance at both the bottom and top ends of the prediction spectrum, and the results from back testing and forward testing while the model runs in production. So the broader takeaway is that our method for handling risk and our credit oversight really hasn't changed amid all this. As a matter of fact, we're quite pleased to maintain that identical degree of rigor moving ahead. One correction, however, regarding your remark that 12% constitutes a floor — I never described it that way. What I said was that we would remain within that vicinity. So I'd rather not make too firm a commitment there. Still, what we observe is that our approach, which stronger models have helped enable in part, involves deliberate risk expansions that yield higher risk-adjusted margin, and that approach is delivering results — which is what we expect to keep seeing for the foreseeable future.

Okay, Jorge, one more thing that should always be considered is the rising LDR and the impact that LDR has on how the business model works. We've brought this up on several occasions: if you review our balance sheet, it remains extremely unlevered. On Slide 16, you saw how that LDR has moved across the last several quarters. There was substantial expansion in Q1, though a big credit cost followed, largely driven by seasonality. Once we move into Q2, we begin to observe the advantages from much of that expansion and from the optimization of the balance sheet. And a meaningful opportunity lies ahead. Naturally, we'll keep optimizing the balance sheet as our credit portfolio keeps expanding; what that means is simply shifting a great deal of deposits that currently earn CDI or the refi rate into assets with considerably higher yields. And clearly that flows straight into margin and into ROE. So that's a really powerful dynamic also taking place inside the business model.

Operator, could you open the line for Mr. Eduardo Rosman from BTG Pactual.

Congratulations on those results. I'd like to ask Dave something — I believe we saw news not long ago that you've joined OpenAI's Board. It would be wonderful if you could tell us how you think you're able to assist OpenAI, and even more so, how that role could benefit your work here at Nubank.

Of course. Thank you, Ed. Clearly, this isn't an earnings call for OpenAI, so I'll keep my remarks fairly brief. But — but essentially, we've spoken quite candidly, both within the company and publicly, that in our view artificial intelligence represents the biggest technology shift ever seen in our history. It will also rank among the most impactful technological changes for any company in any sector worldwide. So this is a worldwide phenomenon, and an extremely strong one. Companies will undergo major change; and while it's still early, we can already observe it plainly at Nubank. We've talked today about many uses, ranging from credit and underwriting, and even things that touch customers directly, and we're in the midst of a major overhaul throughout our company concerning the way we're deploying AI — placing it at the forefront as a technological force and as something that will effectively deliver an edge; so seen that way, for me, being nearer to an organization such as OpenAI naturally gives very valuable perspective. On a personal level too, I believe it's a terrific chance to help ensure that a few of these outstanding AI firms create something wonderful for humanity, and I hold enormous admiration for the OpenAI team and the manner in which they pursue this mission. So I consider it a win-win. And plainly, I mean, it's still early, yet I'm hopeful that I'll be able to add meaningfully to how the organization carries out its work.

Operator, could you please open the line for Mr. Pedro Leduc from Itau BBA.

Thank you, everyone. I have two questions. First one is a bit more of a technical exercise. That portfolio you currently hold with clients under the government renegotiation program that showed up in Stage 3, possibly within your personal loan book. And I would guess there's somewhat lower expected default loss, considering the decline and the coverage — that's the technical part, since I'm attempting to read the movements here. And then the — and the second one, shifting somewhat back to business matters. During the prepared remarks, when you discussed unsecured lending, you brought up the pace. You've offered a few comments regarding that. Could you provide a bit more detail, and also 1 on your most recent update concerning how you stand on the payroll, private payroll included.

All right. Well, I'll begin with that technical question. There was a minor effect on expected credit losses, which I noted during my prepared remarks, yet it was not material, under $10 million. And it does show up in Stage 3 of lending, which is where it appears. I do believe we will experience somewhat more of an effect from Desenrola in Q3, though we've already witnessed over 4/5 of that reaching us or helping us in Q2. And I trust that responds to your question.

Regarding the portfolio currently sitting within the program, is it classified as Stage 3, or does it begin at Stage 1?

Correct. So on that — no, it falls under a Stage 3 portfolio.

Yes, absolutely. And when it comes to private payroll, we're picking up speed each month; we're gradually becoming more at ease with this offering. There's been meaningful movement in the way the product is structured for the Brazilian market, how businesses can obtain the collateral, how the systems operate; so, as we've stated on numerous occasions, we identified it as a metric; and as you wagered on moving too fast too soon with a product carrying so many uncertainties. We believe we're nearing a framework that makes a great deal of sense, and we're gaining momentum. And in the end, the provider with the lowest costs and whoever serves the customer best is going to take this market. And so we believe we're positioned extremely well to be among the leading players in this space across the coming 18, 24 months; so we have — and we consider it a positive for the market. The other piece that's likewise beginning to shift a bit is contrary to what most people would assume, the earliest borrowers of these loans were extremely high risk. These weren't the scenarios you'd anticipate from folks refinancing — low-risk individuals refinancing expensive debt. It was in fact very high-risk customers. And so from that angle, it simply didn't make much sense for us to be opening the door for that while we grasp the actual degree of risk. And there's somewhat of a shift in behavior. Customers actually — solid customers actually spotting the chance to refinance then this turns into a far more appealing offering for consumers. We would be delighted to make that trade. There have been plenty of discussions that we're afraid of cannibalizing ourselves so that we don't want it to succeed. We will be extremely glad to be able to refinance every customer who wants lower interest rates for a product that carries that collateral, it would be highly advantageous for us to have greater diversification in our portfolio. It would bring more resilience; it would be less cyclical. So strategically, it's a product that we are very we believe over the long term will be very good, and we are positioned very well to do it. We're simply proceeding at the right speed, and we're becoming more comfortable month by month.

Operator, please unmute the line so Mr. Uri Fernandez with JPMorgan can speak. David, congratulations on the net income that came in above $1 billion, and I have a question about that figure.

I was doing some checking on Banco, which is one of Brazil's top banks. Looking at its retail business, that was roughly $1.1 billion, correct? So you're quite near that figure. My question, David, is how do you keep pushing this net income higher? I realize your ROE is better, your cost-to-income ratio is stronger, and you've got that efficiency advantage working for you, but the company is becoming really large, isn't it? So could you help us figure out whether this is simply an RPA normalization with your mature cohorts arriving, or whether it's driven by new products? Or does Mexico play into it? Let me congratulate you on reaching $1 billion, but I'd also like to ask how you sustain growth in profitability at this size?