Upstart Holdings, Inc. (UPST) Q2 2026 Earnings Call Transcript

Good afternoon, and welcome to the Upstart Second Quarter 2026 Earnings Call. As a reminder, this conference call is being recorded. I would now like to turn the call over to Sonya Banerjee, Head of Investor Relations. Sonya, please go ahead.

Thank you. Welcome to the Upstart earnings call for the second quarter of 2026. Joining me today are Paul Gu, our Co-Founder and CEO; and Andrea Blankmeyer, our CFO. During today's call, we will make forward-looking statements, which include statements about our outlook and business strategy. These statements are based on our expectations and beliefs as of today, which are subject to a variety of risks, uncertainties and assumptions, and should not be viewed as a guarantee of future performance. Actual results may differ materially as a result of various risk factors that have been described in our SEC filings. We assume no obligation to update any forward-looking statements as a result of new information or future events, except as required by law. Our discussion will include non-GAAP financial measures, which are not a substitute for our GAAP results. Reconciliations of our historical GAAP to non-GAAP results can be found in our earnings materials, which are available on our IR website. For the first time this quarter, we'll discuss contribution margin separately for our unsecured and secured businesses. As a reminder, Upstart has one reportable segment, Unsecured Lending, formerly called Personal Lending. The name change is administrative only and does not affect the underlying disclosures. Our Auto and Home businesses are not separate reportable segments, but in certain earnings materials we refer to them collectively as secured products, which is derived by subtracting Unsecured Lending from total company results. With that, Paul, over to you.

Thanks, Sonya, and thank you, everyone, for joining us today. At the end of our last earnings call, I shared 4 key commitments and takeaways. I want to start back there today, right where I left off. First, I said that core personal loans are a superpower. Our technology lead there gives us unusually strong margins, and I told you we would reaccelerate its growth. Second, I said Home and Auto had found their fit with the market, and I told you we would turn their focus to improving profitability. Third, I told you that we'd stay capital efficient, even as we pursue the enormous opportunity in credit. And putting those together, I told you that we'd drive a rebound in profitability that would show we are on track for our full year guidance. Today, I'm pleased to report that we've executed exactly that plan. We grew core personal loan originations 27% quarter-on-quarter. That's a $526 million sequential increase, which is more than 3.5x the growth of the prior 3 quarters combined. It's also the lion's share of the growth in our Unsecured Lending segment. Because of our strength in this market, we achieved this re-acceleration while also driving our unsecured contribution margin up 6 percentage points compared to last quarter. At the same time, our secured products, Home and Auto, advanced rapidly towards profitability. Combined, their contribution margin improved by 61 percentage points in a single quarter, closing a large part of the remaining gap to breakeven, and they did so while still managing origination growth of 45% quarter-over-quarter. Combined, originations across all products grew 23% sequentially or $782 million compared to last quarter. Our third-party funding strategy delivered at an almost equally rapid pace, allowing us to support that growth without additional equity capital. While loans on our balance sheet increased marginally, they declined to just 5.9% of total outstanding loans, our lowest level in almost 2 years. Our strategy drove a rebound in our overall margins and profitability, including all-time high contribution profit and a return to GAAP profits. To put that in perspective, our previous peak in contribution profit was in Q4 of 2021. Back then, the business benefited from a much easier macroeconomic backdrop and the financial profile of being concentrated almost entirely in a single mature product. UMI was below 1, consumer charge-off rates were at historic lows across the industry, and the federal funds rate was near 0. The fact that we've reached a new profit peak in today's environment is a testament to the relentless power of compounding technology wins quarter-after-quarter, year-after-year. I always tell people that while our business is sensitive to macro conditions in the short term, its value in the long term will be determined only by the pace of our execution. Also, unlike 2021, we are now a multiproduct company. Our business today has a significant and growing share of secured products. Auto and Home made up about 14% of total originations in Q2, compared to just 1% back then. Building those businesses is showing up in our operating expenses now, ahead of the contribution profit we expect them to generate in the future. These products expand our market opportunity by many multiples and, combined with core personal loans, can fuel profit growth for years to come. Next, I want to highlight our progress in 3 areas: our models, our customer relationships, and our secured products. As always, our most important business lever, especially in core personal loans, is improving our models. In Q2, we shipped 3 new personal loan underwriting models, cumulatively adding more than 300 new variables. We also moved personal loan underwriting to a new distributed inference platform that is roughly 65% faster at the median relative to the prior architecture, even as it supports much greater complexity. The end result, our model's accuracy lead over a traditional credit scoring benchmark widened again this quarter. Our model is now 2.74x as accurate as a traditional model, and we're still early. 87.38% of the inaccuracy gap is left for us to solve. That's our runway. Turning to our customers. Q2 was another step towards becoming the most trusted brand in consumer credit. Approximately 1 in every 13 American adults has an Upstart account, and that number continues to grow. Investing in these relationships is important to us. In Q2, we originated more than 558,000 loans, a record high. Historical experience shows that each of those borrowers will take out roughly 1.5 loans over time. Recent cohorts are trending even stronger as the addition of new products like Home, Auto and Cash Line brings us closer to our vision of having the best products for every American's credit needs. We're also making it easier for consumers to return. In Q2, we launched a new model that allows us to better manage when we pull underwriting data from vendors, allowing us to re-engage existing accounts more frequently and at lower cost. Next, I'll talk about our secured products. In Home, we streamlined the borrower verification and closing processes. Our cost to originate a HELOC decreased 15% versus Q1, and we can close in 6 days while also offering borrowers rates that are on average more than 200 basis points lower than competitors. That combination, lower cost, speed to close and better pricing, is the basis for a durable competitive advantage that should support our continued growth in this market. In Auto retail, we continue to add rooftops and win wallet share. At the same time, because of the great value proposition we offer both dealers and car buyers, we began optimizing our take rates, a clear sign this business has moved from proving demand to improving unit economics. In Auto secured personal loans, we improved the efficiency of the funnel and upgraded our ability to automatically identify consumers with eligible vehicles, which lowers acquisition costs and directly supports product margins. Each of these secured products has the potential to be as important to Upstart as personal loans over time, and we're excited to continue investing in them. At the same time, capital discipline means holding an extraordinarily high bar for investments. And because of that, we decided to sunset our Auto Refinance business this quarter. While we're proud of what the team built over the past few years, it did not have the same velocity or potential as the other bets in our portfolio. Turning to funding. Since our May earnings call, we've closed 3 major institutional deals, including our largest ever, which together provide up to $5 billion in new committed capacity. We've also kept our streak intact, renewing every institutional capital partner at a 100% rate since 2023. Separately, we completed an upsized $569 million asset-backed securitization, our largest issuance since 2021 at the tightest spreads we've seen in 3 years. This activity is another vote of confidence in our ability to deliver strong returns to our capital partners. The average return of our last 12 quarterly vintages of loans exceeds U.S. treasuries by approximately 660 basis points, with every individual vintage exceeding treasuries by at least 425 basis points. Finally, a quick update on our bank charter. In July, we received conditional approval from the OCC following a rigorous review of our credit compliance and business practices. This process, plus the work remaining to receive regulatory approvals and stand up the bank, is one of the largest undertakings in Upstart's history. The bank does not change our strategy of funding loans primarily with third-party capital, but we expect it to unlock major operational and regulatory efficiencies, which will contribute to our financial goals over the coming years. We aim to launch in early 2027. Before I turn the call over to Andrea, I'll close with a few final thoughts. Q2 was our first quarter executing with a new management team. We defined a strategy and we executed it. At AI Day last year, I told you that lending's oldest truism assumes the technology stays constant, that you can't have growth, credit performance and profitability all at once. That's not the case for Upstart. This quarter, we delivered all 3. We grew, our credit performed, and we expanded margins. We didn't have to trade one for another, and that combination, not any single metric in isolation, is the clearest evidence that our AI advantage is real and compounding. You should expect us to double down in the second half of this year. We expect to compound wins across technology and marketing as we drive growth in core personal loans and profitability in secured loans. That's where the real durable value in this business lies, independent of any macro headwinds or tailwinds. And we will continue to steward every dollar of investor capital, expanding and deepening our third-party capital relationships and holding a high bar for operational investment. I want to close by thanking everyone at Upstart for an exceptional quarter. Andrea, over to you.

Thanks, Paul, and good afternoon, everyone. As Sonya noted earlier, we renamed our sole reportable segment to Unsecured Lending this quarter, a naming change only with no impact to the underlying disclosures. As a quick reminder before I walk through the numbers, Unsecured Lending includes personal loans, small dollar loans and Cash Line, while secured, comprised of Auto and Home, isn't a separate reportable segment. It's derived by subtracting Unsecured Lending from total company results. This framing reflects a real shift in our business. We're no longer a single product company, and our unsecured and secured products are at different levels of maturity and have very different economics today. Our focus on shifting the mix in unsecured towards core means we are leaning into the most profitable part of our business. Secured, on the other hand, is still working towards breakeven contribution margin and has been improving quickly. Breaking these out separately is the clearest way to track both stories: the earnings power of our more mature unsecured segment and the additional profit engine we're building with our secured products. Turning to the quarter. Q2 was shaped by execution on the priorities we communicated last quarter and the numbers back it up directly: a re-acceleration in core personal loans, a step change in secured product contribution margin and a rebound in total company margins and profitability. I'll cite both year-over-year and sequential growth as I walk through our results; year-over-year for the long-term trajectory, sequential for how we executed against the plan. Total originations were $4.2 billion, up 50% year-over-year and 23% sequentially. Within this, Unsecured Lending originations grew 38% year-over-year and 20% sequentially, with the latter reflecting a re-acceleration of core personal loan volume growth. At the same time, our secured products continue to scale, with Auto originations up 264% year-over-year and 62% sequentially, while Home grew 139% year-over-year and 14% sequentially. Total revenue was approximately $365 million, up 42% year-over-year and 18% sequentially. Revenue from fees was $348 million, up 45% year-over-year and 26% sequentially. Within that, Unsecured Lending contributed $326 million in revenue from fees, up 38% year-over-year and 23% sequentially. Secured products contributed $22 million, up 465% year-over-year and 86% sequentially. Take rate, defined as revenue from fees as a percentage of total originations, improved sequentially in both categories of products, about 24 basis points in unsecured and 81 basis points in secured. That's why fee revenue growth outpaced origination growth versus Q1. Net interest income and fair value adjustments totaled approximately $17 million, roughly flat year-on-year, but down sequentially, reflecting the impact of a higher UMI on fair value. Next, contribution profit, a non-GAAP metric defined as revenue from fees minus variable costs for borrower acquisition, verification and servicing. Contribution profit was $193 million in Q2, an all-time high for Upstart. That's up 37% year-over-year and up 41% or $56 million relative to Q1. The sequential increase was almost entirely driven by Unsecured Lending, with secured products representing less of a drag versus Q1. Contribution margin was 55% versus 58% in Q2 2025 and 50% in Q1 2026. The 5-point improvement versus Q1 was driven by margin gains in both our unsecured and secured products. Unsecured segment contribution margin increased to 62%, up 6 percentage points from 56% in Q1 and flat to Q2 2025, with the sequential improvement reflecting 3 things: one, a larger mix of higher-margin core personal loans; two, lower customer acquisition costs as a percentage of originations; and three, an expected seasonal pickup in demand. The re-acceleration in core personal loan volume was driven by a combination of model improvements, funnel improvements and efficient targeted customer acquisition, all reflecting our increased focus on the borrower category. Our secured products contribution margin increased to negative 35%, an improvement of 61 percentage points from negative 96% in Q1. This was driven by improved take rates and a greater operational efficiency across Auto and Home. Given this trajectory, we expect our secured products to reach contribution margin breakeven by Q4 of this year. In total, GAAP operating expenses were roughly $350 million in Q2, up 39% year-on-year and 11% sequentially. Variable expenses, comprised of borrower acquisition, verification and servicing costs, rose 55% year-on-year and 11% sequentially. Fixed expenses, defined as total operating expenses minus variable expenses, increased 28% year-over-year and roughly $19 million, or 11% sequentially. Looking ahead, we expect fixed expenses to grow in the low single digits sequentially in both Q3 and Q4. In Q2, we returned to GAAP profitability, generating approximately $17 million of net income, up 195% year-over-year, with a 5% net income margin. GAAP diluted EPS was $0.16, based on a weighted average diluted share count of 110 million. Adjusted EBITDA was approximately $77 million, up 45% year-over-year, with a 21% margin. We ended Q2 with approximately $1.06 billion in loans held on our balance sheet, up approximately $50 million or 5% from Q1. That increase was driven by our secured products, which continued to scale quickly. At the same time, our unsecured holdings declined and legacy securitized loans continued to run off. As a percentage of the total unpaid principal balance of all Upstart loans outstanding, loans on the balance sheet fell to roughly 5.9%, the lowest it's been in almost 2 years. Supported by consistent credit performance, we've continued to strengthen our capital platform. Year-to-date, we've signed committed capital partnerships that are expected to add up to $10.8 billion in incremental capacity. We also completed 3 securitizations for roughly $1.7 billion in total collateral and increased the proportion of Home and Auto loans funded by third parties. Looking ahead, we're reiterating our full year guidance. Total revenue of approximately $1.4 billion, fee revenue of approximately $1.3 billion and adjusted EBITDA of approximately $294 million, or roughly 21% of total revenue. Keep in mind, our guidance is informed by our most recent published read for UMI, which as of yesterday was 1.5, up 9% from the beginning of Q2 and at the top of the 1.4 to 1.5 range that framed our outlook when we initially shared our 2026 guidance in February. With UMI having trended higher over each of the last 3 months and now at the top of the guidance range, we are maintaining our guide. We expect the underlying strength of the business, as you saw in Q2, to offset this macro headwind. Our outlook assumes UMI holds roughly at this level through the rest of the year. To close, in Q2, we did what we said we were going to do, demonstrating that we could drive sequential improvement in contribution and overall profit margins by: one, reaccelerating the growth of core personal loans at an efficient customer acquisition cost; two, meaningfully improving the contribution margin profile of our secured products while maintaining strong growth; and three, managing fixed expenses. Along with preparing to launch Upstart Bank, these 3 areas remain our focus for the balance of 2026. And above all else, we will continue to prioritize consistent credit performance. Whatever the UMI context, if we execute across these domains, our platform will be stronger than ever as we exit 2026. With that, I'd like to turn it over to the operator to begin Q&A.

And the first question will come from Kyle Peterson with Needham.

Nice results. I want to start out and dig a little more into the take rate. Great to see the uplift there. It sounds like it was across products. So I guess I just wanted to dig in, is there any mix at play there, whether it's in terms of where in the credit box some of these loans were originated? Or is there anything pricing that you guys did to tweak that? Any more color there would be great.

Yes. Thanks for the question, Kyle. We're really pleased with the results on profitability this quarter. And it's really a reflection of the strategy that we laid out at the end of last quarter's call. We said that there are two important dynamics that are going to be happening through the rest of this year. One is we're going to be focused on re-acceleration growth in this core personal loan segment. That segment carries much higher margins. And because you can see in our results that that segment grew much more than it's been growing in earlier quarters, that is contributing to higher margin results. And then on secured products, we very intentionally were focused on actually improving significantly the margins of Home and Auto, and we did exactly that in this quarter. That was our number one goal for those products. And so you can see that the margins improved very substantially, 61 points in a single quarter. So the answer is a bit of both. We're very focused on our core personal loan segment, where we're very strong, but also each of the underlying businesses did really well on their margins too.

Great. That's really helpful. And then maybe switching gears, I wanted to talk about some of the secured products scaling and particularly on potential distribution partnerships. I know some competitors in the space, especially in HELOCs, have used some of these as a way to really ramp up growth, partnering with mortgage brokers or other channels. How are you guys evaluating and incorporating these into the growth strategy of some of these secured products? What do you guys have now and where do you see that going moving forward?

Yes. Each of our Home and Auto businesses have some differences in the distribution strategy. So you're absolutely right that in the Home business, we think ultimately it will be very important to have Home-specific partnerships. That's not something we've done yet, but it's very much on our roadmap. It's one of the areas that we want to invest in. We think there's a lot of potential because our HELOC product is strong in terms of best-in-class prices we can offer borrowers plus a best-in-class experience. Putting those together delivers an exceptional product, and distribution will take it the distance. In our Auto product, there are two different Auto products. There's an Auto purchase product, and that's distributed at car dealerships via Upstart's proprietary software. That strategy has been ramping nicely, as you can see in the results. One of the benefits of that approach is that, unlike our pure consumer businesses, in addition to improving models and user experience, the Auto purchase business can grow by getting to more car dealerships. There are thousands of car dealerships in America that we aspire to scale to. We also have Auto secured personal loans that are more like our traditional personal loan business in being direct-to-consumer. So each product has a distinct distribution strategy; some are already in play and some, like HELOC partnerships, are still to come.

And the next question will come from Simon Clinch with Rothschild & Company.

Yes, nice quarter. I was wondering, Paul, could you talk a bit about the Auto refinance business that you're sunsetting? I'm curious because that strikes me as a business that would actually be quite important for repeat business, generally speaking. So I'm wondering if you could put a bit of color around how that fitted into your strategy and why it doesn't necessarily belong there, aside from the economics.

Yes. We think and we thought that Auto refi is a good product. I think we built a good product over the last few years. And to your point, it was something that was relevant to returning customers. But ultimately, we looked at the growth rate of that business and its potential compared to everything else that we were doing and the other bets we had out there, and we wanted to concentrate on the ones that had the highest velocity and the biggest upside. So we decided to concentrate a little more, and that one didn't make the cut. It was a necessary consequence of capital discipline.

Okay. Understood. And then maybe, Paul, as we see the improvements you're continuing to drive within the secured lending margins, could you talk about some of the low-hanging fruit that's left to be taken and how we should think about the future profitability of this business? Any color on that would be useful.

Yes. There's still a lot of well-defined work to be done on improving the profitability of the secured products. Broadly, they fall into two buckets. The first is improving and optimizing where we take our economics. For example, in the Auto purchase business, there is an enormous variation in how sensitive any particular customer or car purchase deal is to the take rate. In some deals, we offer an extremely unique value proposition and that car would not get sold otherwise. In other deals, we're competing in a more fiercely competitive market. Take rates can be optimized significantly. Over time, as our separation and our ability to underwrite the underlying credit grows and as we better understand which deals are truly competitive versus where we're adding value, that will improve our economics. The second category is making it more efficient to originate these loans. In secured products, there's more cost involved in originating the loan: verification, liens, and all the process steps. There's room for data integrations, automations and more sophisticated AI to automate much of the work. These efforts have short-term components that can hit fast and longer-term benefits where the product differentiation and value continue to grow.

And we'll take a question from Will Nance with Goldman Sachs.

I was wondering if you could talk a little bit about take rate dynamics on the Personal side. It seems there were mix shift dynamics between positive seasonality this quarter, the mix shift towards core personal, and pricing investments. How would you frame the puts and takes on that line item from here, given the nice improvement sequentially this quarter?

Sure. You nailed the drivers of the improvement in take rate on the Unsecured segment quarter-on-quarter: primarily the mix shift to core, where we have a higher take rate on average, and some seasonal lift since Q1 tends to be softer and have lower takes. All else is relatively consistent quarter-on-quarter. Looking out, expect relative consistency with where we are on take rates today. We'll continue to focus on driving growth in our core personal loan segment, which has strong take rates, but ultimately we are optimizing for high-quality revenue and contribution profit dollars to the platform, not maximizing take rate itself.

Got it. That's very helpful. If I could squeeze in another one on OpEx. With OpEx up almost 30% year-over-year, can you talk about how you're thinking about the growth algorithm there and incremental margins over a longer period?

Absolutely. Part of the fixed cost growth this year reflects investments in newer products and secured products as they drive toward contribution margin profitability. We made some investments in Q1 and Q2 related to tech and model infrastructure and preparing for the bank. We also had a one-time restructuring with severance expense this quarter. Many of the investments to support the business were made in the first part of the year, which is why we expect fixed OpEx to grow at a moderate, low single-digit rate going forward through the remainder of the year, which should deliver real operating leverage this year and in the outer years.

And the next question will come from Dan Dolev with Mizuho.

Great to see those results. Congrats. Two quick questions. First, on your conversion rate, it has increased to 19.7%. Just wanted to know the DNA of your new borrower in terms of FICO. What can you tell us about it? I have a quick follow-up after that.

We don't tend to think a lot about our borrower in terms of FICO scores for two reasons. First, our core DNA is that FICO is one of the things we think can be improved in how borrowers are understood, and that's what we work on every day. Second, we've expanded significantly in products, use cases and the range of consumers we serve. We aim to have the best credit product for every type of credit need. We're serving a full spectrum of people: those new to credit or repairing credit, all the way to prime borrowers. Because of that diversity, the conversion rate metric is very sensitive to mix and is a little hard to interpret, which is why we're planning to replace and sunset that particular metric. That said, this quarter we focused on re-accelerating core personal loans, which historically tend to serve borrowers in the medium FICO regions, and that product received emphasis this quarter.

Great. And my quick follow-up: on the guide, amazing results and you're not raising the guide. Should we read that as simply being conservative?

We are pleased with Q2 results and see strong underlying business performance. UMI is at about 1.5 as of the print yesterday, which is at the high end of the 1.4 to 1.5 range we set when we issued guidance. That represents a modest headwind on originations and our fair value marks. Taking all of that into account is the context in which we're maintaining our full year guide.

And we'll take a question from Peter Christiansen with Citi.

Andrea, I was wondering if you could interpret some of your previous comments on UMI a little more, help us understand. You mentioned the base business is still doing well. Is that a function of some of the new third-party capital that you brought in or some of the new product areas? What's giving you confidence about maintaining the outlook despite UMI moving higher?

Pete, it's really all of the above. Our core personal loans business is the centerpiece and is driving much of the financial results we saw in Q2 and will continue to drive results for the rest of the year. We have been executing at a high level against priorities. We're facing a modest macro headwind with UMI at 1.5, which affects originations and fair value marks. But we expect the personal loans numbers to matter a lot. We also expect the secured products to continue improving towards contribution profitability this year, which will help. But we can't achieve the numbers without core personal loans continuing to perform well.

Fair. It looks like you talked about repeat borrowers being up. Could you take us through that dynamic and to what degree that's contributing to lower borrower acquisition and verification cost?

That trend has been playing out over many years, not a single sharp change in Q2. In the early years, Upstart had a single product and was more transactional. Over time, as we've rolled out new products that serve people in different life stages—Auto, Home, Cash Line, small dollar—these give us relevance over a longer period and the opportunity for more repeat business. We've also invested in the experience of coming back to Upstart, making it faster and easier, which continues to be a focus. The net result is that the average loans per borrower metric, historically around 1.5, is trending above that for more recent cohorts, and we expect this trend to continue as we add products and improve the experience.

And the next question comes from Mihir Bhatia with Bank of America.

I want to zoom in on the EBITDA guide between Q3 and Q4. Your guidance implies about 24%–25% EBITDA margin in the back half of the year and you've mentioned secured breaks even in Q4. With fixed costs increasing single digits, could you talk about the cadence between Q3 and Q4? Is there seasonality or should we expect step function increases each quarter?

We told you last quarter that the guide was back half weighted. We're continuing to see that here with EBITDA expected to ramp in the second half. The trajectory from 13% EBITDA margin to 21% this quarter versus last shows strong progress. Looking ahead, expect continued progress on the same levers: growing core personal loans, improving contribution profile and unit economics of secured products—expecting secured to be contribution profit positive in Q4—and disciplined fixed cost growth in the low single digits. That combination should drive the back-half cadence.

And on competitive intensity in core: you clearly re-engaged this quarter with better growth. Can you talk about what you're seeing from competitors? Is your pricing power expanding or holding? Anything in Q2 that gives you more confidence the moat is widening?

We have said our advantage in core personal loans is significant and durable, driven by the differentiation we've built over a decade-plus in modeling credit risk. This quarter made it clearer: we grew core personal loans while improving margins and reducing customer acquisition cost. That combination is only possible with differentiated technology. The primary difference this quarter versus earlier ones is management focus: we were extremely focused on the core segment and it showed up in the results. We believe that advantage can continue.

And moving on to John Hecht with Jefferies.

I appreciate the broken-out detail in secured versus unsecured. First, thinking about the next couple of years, what would you like to see in terms of the mix of funding from forward flow versus ABS? Do you have a balance in mind that you think is optimal?

A key progress over the last few years is shifting our capital base toward committed capital partnerships. Today, well over half of our capital is from committed partnerships. We've been renewing partners and increasing commitments. That's the most important factor for us: a stable, resilient funding base with visibility across months and ideally multiyear horizons. Over the 1–3 year horizon, we'll continue to scale that capital base sustainably and find the right balance between at-will securitization and committed capital, focusing on committed capital for visibility.

Okay. And a separate question on the secured products: in core your turnaround time is quick due to tech. What is the turnaround time for a HELOC or an Auto loan compared to industry standards? How much more room do you have to improve?

Today, they're far slower than personal loans, but we believe they are industry-leading. In HELOC, for example, we're closing in about 6 days, while banks and credit unions can take weeks or months. There's still a lot of room to improve these processes with automation and integration, which will reduce cost and speed up the experience. That's a high focus and we expect more wins the rest of this year.

And the next question will come from Giuliano Bologna with Compass Point.

Congrats on the great results. There's a new disclosure for loan sale fees. I'm curious if that's related to forward flow deals for HELOC and Auto or something else?

Thanks, Giuliano. Yes, those are primarily related to our secured products and fees associated with forward flow arrangements. Historically they were immaterial but reached materiality in Q2, so we've pulled them out as a separate line item.

Going back to the guide and commentary around expenses moving higher: is there an expectation of a step-up in stock-based comp continuing after 2Q? Is that a lever that should flow through as part of the step-up in expenses?

In Q2, about half of the fixed expense growth quarter-on-quarter—about half of the $19 million—was driven by an increase in stock-based comp. About $4 million of that $9 million increase was discrete to the quarter and is not expected to repeat in Q3 and Q4. About $4 million relates to the full quarterization of performance-based RSUs granted in March. So roughly $7–8 million of the increase is attributable to those two factors. We don't expect those to repeat quarter-on-quarter going forward. For the full year, we expect SBC to be about $170 million, representing a percent of fixed expenses in the low to mid-20s and a modest reduction versus last year.

That's very helpful. One brief one: there's been a little increase in Auto and Home loans on the balance sheet. Where do things stand in terms of securing final agreements to continue to execute a lot of those loans off the balance sheet going forward?

Funding progress has been strong. As Andrea mentioned, we've added over $5 billion of committed capital in the last quarter alone. Some of the new deals are flowing into Home and Auto. Look at originations versus how much landed on the balance sheet; the math implies we're adding a lot of third-party capital for both core and new businesses, and we're excited about where that's headed.

And the next question comes from James Faucette with Morgan Stanley.

I wanted to ask one operational question and then one on lending philosophy. On operations, I noticed your fully automated origination percentage decreased slightly by about 100 basis points from a high level at 92%. Does that imply we're bumping up against the top of that range, or can you push it higher to drive better profitability?

No, I don't think we're at the limits of automation. There's a lot of mix effects under the hood across products. For unsecured products like personal loans, automated loans convert at a much higher rate than nonautomated ones. Currently 77% of applications are approved automatically, so that number can still go up. For new products like Home and Auto, automation levels are significantly lower, and there's much more room for improvement. As those products scale and automation rises, they'll contribute materially to profitability.

Got it. Philosophically, following up on UMI: as you're lending with a higher UMI score, how do you validate that you're getting the expected performance from the loans in terms of payback rates or delinquencies? How should we sensitize to UMI moves and their impact on your willingness to lend?

Measuring and predicting credit performance is what we do every day. UMI is up over the last few months and we've invested a lot in that system. We believe we have the fastest and most precise understanding of macro effects in consumer credit. Under the hood there's sophistication to understand what is happening with different segments and combinations of characteristics. That deep learning-based approach lets models react to the latest data and price risk into underwriting. We expect properly calibrated performance, meaning loans that deliver returns similar to our targets. The renewal rates and committed capital growth show our partners are happy with the returns.

And we'll take a question from David Scharf with Citizens Capital Markets.

Congrats on strong results. On the capital side: on an absolute dollar basis, it looks like the amount of core personal loans retained on the balance sheet is going down. As the product is more mature and higher margin, should investors think about a timeline or a target where the company feels it doesn't have to tie up capital in retaining any of the personal loan product?

When we think about balance sheet use, currently about 5.9% of total principal outstanding is retained on our balance sheet. The balance sheet serves two main purposes: R&D and timing of loan sales and aggregations. We're doing less R&D as a percentage of originations versus a year ago, but we still have R&D across product categories, including Personal Lending and Auto and Home. We also use the balance sheet for timing of loan sales that may cross month-ends or quarter-ends. So there's no absolute target to bring the balance sheet to zero; as long as we have a strong capital and liquidity position, we're comfortable using the balance sheet for those strategic purposes.

Got it, that's helpful. Follow-up: regarding co-invested capital—about $1.3 billion—besides the qualitative alignment benefits, how do you analyze co-investing capital versus lower loan sale prices without co-investing? Other fintech lenders often have whole-loan sales without co-investment. How do you quantify the advantage of committing some capital?

We don't view co-investment as similar to selling loans at a discount. We expect to earn returns on co-invested loans. The essential calculus for using co-invest is to lock in longer commitments from partners. These deals can be committed for 12 to 24 months, providing predictability in our funding supply and insulating us from market shocks. From a partner's perspective, committing multiyear requires comfort about how we'll underwrite over time; skin in the game addresses that. The number and size of deals, and their length show this structure is working. We view it as a beneficial trade: predictability and commitment from partners in exchange for a modest piece of skin in the game, from which we also expect positive returns.

And we'll take a question from Rob Wildhack with Autonomous Research.

A question on the July volume number that came out last week. Originations were up 50% in the second quarter, pretty consistent month-to-month, and July drops to 34% growth. I don't think there's a comp issue versus July of last year. UMI is up a little but not a ton. Is there anything to call out with that slowdown, and any additional context on volume growth for the remainder of the year?

We saw a modest step down in July versus June, and June was flat to May in originations. That trend is partially reflective of the UMI context across Q2; UMI rose about 9% from the beginning to the end of Q2. The July results include some of that modest headwind from the uptick in UMI. As we've said, UMI moves create headwinds or tailwinds for originations. Over the remainder of the year, we're focused on levers we control: model and technology improvements, customer experience, and marketing efficiency. Those are the key drivers we've used historically and expect to continue using. UMI will move in the background, but our focus remains on controllable drivers of sustainable growth.

Okay. One more on the bank and the charter, now that you're accumulating approvals. I understand the operational benefits. Could you shed light on how you'll run the bank once it's operating: how much capital to seed it, how quickly you plan to start originating loans through the bank and what portion you might do, and how quickly you plan to scale deposits? Any details would be great.

On capital specifics, we haven't disclosed the detailed capital plan, but we are sufficiently capitalized today to launch the bank. On timing and operations, we expect to move the bulk of our originations to Upstart Bank relatively quickly after launch from current partners, and we expect to be able to raise deposits within a relatively short period after bank launch. Overall, we expect the core elements of bank operations to be up and running relatively shortly after launch.

And I will now turn the conference back over to Sonya Banerjee.

Thank you. For the first time this quarter, we invited retail investors to submit questions through X. We received a number of thoughtful submissions, and we've selected a few themes to answer today. So Paul, starting with the first one, you've talked a lot about LTV efforts and have stated Upstart is deliberately not maximizing take rate today. As the models improve, how much pricing power is accruing that you're choosing not to harvest? How do you think about pulling the lever on take rate over the long term?

It's a great question. Over the last couple of quarters, one of the themes we've emphasized is investing in our customer relationships. We expect to have products that serve customers across their life cycles. Getting customers into our ecosystem, even if they don't take out a loan immediately, is extremely valuable and will pay dividends over time. We don't want to over-monetize consumers or maximize the margin on every transaction. Even though in some products, like core personal loans, our pricing power is large and we could push rates higher, that's not our strategy today. We want to take a healthy margin reflective of the value we add, but leave plenty on the table for consumers so that value accrues to our brand and relationships. We did not achieve Q2 results by squeezing take rates in core. If anything, we'll look to invest more in customer relationships in coming quarters rather than extract more pricing.

Thank you. Next question: you say the model is at an accuracy that is 2.74x a traditional model and that 87.38% of the inaccuracy gap remains. How much of that residual error is irreducible? Isn't there a natural limit to prediction accuracy, and how do you know you're not close to it?

Physics is real, so there are limits to predictability. But the starting point for industry models has been a relatively low bar. Traditional credit scorecards and human underwriting solved only a small fraction of the inaccuracy. We've improved to 2.74x that benchmark, but a lot of room remains because the starting point was so inaccurate. We haven't observed clear diminishing returns from further investments in data and modeling; each improvement keeps increasing separation. We're confident that in the foreseeable future our models will keep improving and widening the lead, enabling further differentiation and better value for customers.

Great. Final retail question: what do you believe the market is still missing about Upstart? What specific milestones should shareholders watch over the next 12 months that could help close the valuation gap?

There's a gap between how we see the business and how the market sees it. From our perspective, the business is stronger than ever: better tech, larger committed capital, more customers, traction in Home and Auto. These are wins that add up regardless of macro. Consumer credit is massive, and AI is well-suited to transform it. We've worked on this problem for over a decade and have a head start. We believe the market has been focused on near-term questions about growth, profitability and funding. We need to continue proving that we can grow profitably and fund the loans we originate. If we address those doubts through execution and deliver profit improvement, then across valuation frameworks, profit growth will drive long-term value. Ultimately, if the market is skeptical, we can earn the profits to invest further and compound value ourselves.

Thank you. And that does conclude today's conference. We do thank you for your participation, and have an excellent day.