Upstart Holdings, Inc. (UPST) Earnings Call Transcript & Summary

Welcome, everyone. I'm Henry, better known as Henry Invests on Twitter. Today, I'm really excited to be joined by Upstart's CFO, Sanjay Datta, who is here to answer questions submitted by retail investors. There will be a live portion of this event for questions. So if you do have a few that pop up, feel free to put them in the chat, and we'll take a look at them in just a little bit. About 6 months ago, Sanjay and I had our first Q&A in August, which really set off the start of the Upstart comeback story, which has continued to progress and gain momentum. So I'm excited to learn a little bit more about that today. So Sanjay, thank you for being here, and it's great to see you again.

Great to see you, Henry. Thanks for having me.

No problem. We're really about like 3 quarters into the Upstart comeback story. Curious if you could share a little bit about really how that's materialized over the past year and maybe what you're most excited about for the future.

Yes. I mean I attribute it to the chat we had. That kicked everything off, didn't it? No. I guess at a high level, the environment got really bad for a while. I think that obfuscated a lot of the good work that I think we're doing internally. And then it's as simple as like it stopped getting bad. It stopped getting bad probably a couple of quarters ago. And as a result, I think a lot of the good work that's been happening internally is starting to sort of bear fruit. Along the way, we -- I think it was probably realized that there's a lot of things in our business that could be sort of refined or retooled, and so we took the opportunity to do that. And now that the environment is not as stiff of a headwind, I think that you'll see a lot of that stuff start to materialize as good business performance. So I think that's what you're starting to see in the last few quarters.

Yes, we've seen a lot of micro improvements. In Q4, you launched Model 19, which has a new feature called the Payment Transition Model. And you also launched a new innovation called Aqueduct. Can you tell us a little bit more about these new AI and model advancements?

Yes. Sure. That's sort of the bread and butter of what we do. Maybe on Aqueduct, Aqueduct is a bit of more of an engineering productivity sort of innovation. But I just more generally, we have an ensemble of approximately 30 models now that they're all trying to refine predictions about various parts of the process of originating and servicing a loan, and they're constantly improving whether through new data sets or better algorithms over time. So yes, we talked about Model 19 was an example. It's just an anecdotal example of that. The PTM model that you described essentially means -- so prior to this launch, the model essentially had a more limited understanding of the final state of a model. It was either current or charged off, and it knew if it charged off roughly when. So it had an understanding of timing. But of course, there's a lot of states in between that, you can have a model that's not current, but not yet charged off in delinquency and models in delinquency can take various paths through space time, right? You can sort of roll the charge-off or you can recover. And Model 19 has a more nuanced understanding of all those different pounds. And so when it looks at a training data set and there's a bunch of delinquent loans, it has a better understanding of which ones are more likely to roll to charge-off and which ones are not. So all that is a way of saying it's just a smarter way of assessing at a loan-specific level what we think the outcome is probable to be. And that may sound like a very nuanced thing, but I think you add a lot of those up and you start to get much better predictions about the loan outcomes. And so that was just an example of one of the things that it's really the core of what we do day in, day out.

Yes. I want to talk a little bit about risk sharing. We've seen these risk sharing agreements for Upstart pick up throughout 2024. They're really often cited as sort of a bearish thesis. Can you share a little bit more about how Upstart is thinking about these agreements? Do you plan to continue to increase them? And maybe what causes the value of these agreements to really fluctuate?

Yes, of course. Yes, I think that's been a big sort of centerpiece of our more recent strategy. In a sense, you can think of it as a bit of a transition for us. If you sort of rewind a few years back, we would have said something like, we were a purely transactional platform. We don't really have any sort of involvement in the performance of the credit itself. And as a result, the capital that was being used to fund the loans was sort of free to come and go at will. And I think the lessons of the last few years have taught us that it's a better strategy to be a largely transactional business model, but that has a bit of skin in the game in exchange for a capital base that is going to be more resilient and more committed when the world and the market goes through a bit of turbulence. And so I guess the bearish take in a sense would be, well, it's not a purely transactional model. There's some skin in the game that represents some amount of risk in the model. We think it does a lot of good things. It aligns incentives between us and our capital partners. It creates a more resilient model, a less volatile model. And it's a trade that it's well worth it. I think it's a risk that we're set up to manage. So we view it as a very positive development. So we do plan to continue scaling these roughly in proportion to the overall platform. So we always want it to be some percentage of our capital. Yes. And the value of those agreements will fluctuate based on the performance of the underlying credit. Now important to note, that's an aggregate statement, right? So I think that the way we think about our business is any given vintage in a benign period, we would expect it to outperform. There will always be macro shocks in which they underperform. Over the duration of a relationship with a counterparty, we care about the aggregate performance. And if that underperforms, then these values will be marked down. If they overperform, we will reap the vast majority of the overperformance. So these partnerships are designed to perform over a cycle. And I think that it's at that level of aggregation that you could see fluctuations in valuation if there's overperformance or underperformance.

Appreciate the color there. Moving on to auto a little bit, really a strong quarter. Originations grew 59% sequentially with transaction dollars growing 61% sequentially. Curious really to know what is the value proposition that Upstart Auto brings to dealerships? And what is sort of the growth road map look like for that product over the next couple of years? Do you see Auto being potentially meaningful to the top line by, say, 2026?

Yes. It's -- we're -- I think we're feeling pretty gratified as our core business emerges that we took the decision to protect some of the sort of the smaller earlier-stage bets that we were working on through some of the lean times, things like Auto, HELOC and small dollar lending, I think, are very large opportunities. And now that the world has stabilized a little bit, I think you'll start to see those contribute meaningfully. The value prop for Auto is it's sort of the same as in all segments of credit for us, which is that most credit segments have this shape, where there's like some population of very well, highly served customers. And then below some arbitrary credit score cutoff, you sort of rapidly descend, like not well served at all. And our opportunity, we sort of essentially always aim to do 2 things. One is bring more people into the well-served segment at lower APRs. And then we also aim to make the process of accessing the credit much more frictionless and effortless for all borrowers. And so that's true of Auto as well. And I think our plans for Auto in 2025 involve a lot of scale up. It's from a small base, so I don't think they will necessarily be meaningful to the P&L or the financials in 2025. But yes, I think we absolutely hope and expect that this will be a meaningful contributor to the financials by 2026.

Awesome. Yes, another, I guess, product segment that spending question is mortgage lending. In Q4 2021, Upstart had mentioned trying to enter that market by 2023. Obviously, a lot has changed since then. Just curious maybe if that's still a market you're looking to get into in the future or if recent shifts, I guess, in, say, strategy and macro have kind of changed that outlook. I know for me personally, my generation, Gen Z, it's pretty difficult to purchase a home and maybe equally as difficult to finance one. So do you think that Upstart can help with that in the future?

Yes, absolutely. That was the problem we had started down towards a few years back, as you mentioned. When all of the stuff in the macro went down, we downscoped our ambitions. We -- as I said, we didn't -- we wanted to protect some level of continuity in these young bets, but we essentially downscoped it to what's called a home equity loan, which is, in some sense, it's a much lighter lift. It's an easier nut to crack. And so we've made progress there. And even though that itself, we think, is a big exciting opportunity, it is ultimately, I think, probably a stepping stone on the road to purchase mortgage, as you described. And as usual, it's with the underserved population where there's a large opportunity. Like if you go back to, I think, probably like sometime around -- if you go back like 2001, there were like 1 million mortgages a year being originated that did not qualify for Fannie Mae or Freddie Mac, those are the government-backed institutions that back the majority of the well-served sort of segment of the market. So there's a lot of what you'd call non-qualifying mortgages being originated. And then, of course, that got too exuberant by '08 and then it sort of blew up. And then the baby kind of got thrown out with the bathwater a little bit. And now there's like barely any. It's really hard to get a -- if you don't qualify for Fannie or Freddie, it's sort of like hard to get a mortgage with respect to process and good rates. And so as a result, there's a large number of people out there today who are good credit who can't qualify for a reasonably priced mortgage. And I do think we certainly aim to fix that. The exact time line is -- I don't think we've sort of laid that out, but I would definitely assume that that's something we're going to turn our attention to at some point.

Awesome. Sounds good. Shifting, I guess, a little bit towards macro now. Newly elected President Trump is creating headlines as per usual. He's promised to impose tariffs as high as 25% on both Canada and Mexico. He has also suggested capping credit card interest rates. Do either of these actions really affect Upstart? They've affected the broader market in general. And does Upstart really plan to get into credit card underwriting in the future?

We certainly don't wage too far into the exercise of macro prediction. I think it's as difficult as ever to try to predict what's going to happen. Let's see. I mean it's -- tariffs -- neither of those things, I think, will impact us directly. Now tariffs generally are not great for interest rates, and they can lead to inflation. So I think those things you might think of as risks if they materialize. Let's see. Credit card caps. So we don't -- we're not in the credit card space today. In fact, we tend to sit behind credit card debt as a refinance tool when those things are overpriced. So again, a cap wouldn't affect us directly. I think if you placed a cap on credit card APRs, you would effectively see a dramatic lowering of approvals. So there's a bunch of people who would just not get approved for credit cards. And I think then there's probably a large population of people who take credit from somewhere and that would be an opportunity for us. So yes, I think that any impacts of either of those things would probably be like second order impacts. As for whether we -- I think you asked if we expect to be in credit cards on some level. We haven't announced any plans to be in credit cards. I do think that -- maybe just let me generalize that the more general segment of revolving credit, whether on a sort of credit card or in some other instrument. I think it's obviously a large pool of credit, and it's interesting to us. But we don't have any definitive plans there as of right now.

Got you. Yes. I guess another area that a lot of people are interested in is crypto, cryptocurrency. It seems like a lot of companies are picking up and joining this train recently. So I guess my question is, what's sort of Upstart's thinking around crypto? Is that something you plan to incorporate in your business or maybe even as a balance sheet item in the future?

Yes. The short answer is not really, to be honest. We've certainly thought a lot about it. I mean I think we're determined to remain very credit-centric. I think that we believe in the power of focus and then doubling down on what you're good at. So we have done some thinking around how crypto may be involved or adjacent to what we do. And there are some ideas floating around, like to the extent we went international, you could use it as a cross-border payment instrument and sort of simplify the whole currency management regime. You can use it as a security in lending. So there's ideas, but I think they're sort of like a little bit marginal, to be honest. And so we have people who are interested in that rough world, but we haven't found a way to make it very central to what we think we're good at and to what our mission is.

Got you. Yes. Something Upstart is good at, though, is really operating leverage. Do you think -- can you share a little bit more about that? Do you think -- I know you guys want to scale profitably. So can you just describe maybe your current cost structure and how you think you can grow while keeping that maybe relatively stable?

Yes. It's, I think, an important part of our business model. I think we do have a lot of strong operating leverage in the business. I think a lot of analysts out there were questioning whether we could get back to profitability by like 2027 or something. And I think they sort of misunderstand our model a little bit. But a lot of our margin strength derives from the fact that as a mission, as an orientation, we are serving the underserved, right? And as a result, we are in a part of the market where not a lot of others are willing to go. That's the whole point. They're underserved. And if you can serve them effectively, you can have a lot of margin strength because there's not a lot of people -- a lot of other competitors creating adverse selection and competing away margins. And so as a result, I think that as we grow, I think 2 things -- there are 2 things we believe about our P&L. One, as our contribution margins will remain relatively steady. And I think that's one thing that, again, in lending traditionally, when you scale, you erode your contribution margins. And the reason is because, if you don't have underwriting that's improving over time, if your underwriting is more static as it has traditionally been in the industry, then the only real ways to grow are by starting to overspend on marketing for the next borrower or by degrading your credit quality and doing stupid things there. And so most people assume that when you grow, you're going to sort of grow at eroding margins. But if you grow because your underwriting is getting better, meaning over time, you're avoiding a few more of the defaulters. And as a result, you are able to give the remaining good borrowers lower APRs because they're not subsidizing as much default, your conversion rates get better, all the goodness of how we think our business model grows, then you can grow actually with improving acquisition costs. And I think we've demonstrated that in the past. When we were on good growth trajectories back in 2021, we actually grew at declining acquisition costs. So all that to say, I think we have demonstrated and we will continue to demonstrate the ability to scale at sort of steady contribution margins. And then our fixed cost base is very lean. And I don't think we need to add to it dramatically in order to scale our core business. All that to say, I think that money will drop very efficiently to the bottom line as we scale. And as Dave said on the call, if you caught it, we want to be one of those unique companies that combines high growth with strong profitability, and I think we have an opportunity to do that.

Yes, for sure. We've seen a lot of the micro improvements really revealed publicly over the past couple of quarters. What caught me off guard and maybe really excited about the earnings release on Tuesday was really your announcement of Upstart AI Day, which is scheduled for May 14 in New York City, where you're going to expose a little bit more of these data points. Can you maybe give us a teaser or something you're really excited to share about this event?

Yes, we're very excited about this. It's going to be, I think, very tech-centric for us. We're going to want to sort of explore our technology with you guys a little bit more deeply. It's sort of hard to do that on an earnings call or in a Q&A, which are predominant public forums today. But I think we talked about this last time, like if you just think about where the areas of skepticism could be about our business. One of the big ones is like you may believe in AI as a technology and you may believe that it will have a big impact on credit, but how do I know that Upstart is really doing this? Like isn't every sort of digital lending company doing some version of this? And we think what we're doing is very unique, and we want to expose that in a bit of a deeper way. So we'll talk a lot about our technology and why we think it's differentiated and unique and why we think there's really nobody else that is doing something very similar right now in the U.S. credit markets. And we'll talk about some other things that I think are important topics, like I think there's another question out there around like, how big is this opportunity? Like is this just like a little niche thing like, oh, you found a few mispricings on the margin. We, of course, believe that mispricing is right in credit. In fact, it's dominant and that, therefore, if you can have better risk models, you can improve it dramatically across the aggregate market. And so we'll talk a little bit about our framing for why we think the opportunity is as large as it is. And then, of course, there's another topic around volatility of the business model, right? Like, okay, so it's great that you're growing right now, but what do you look like in the next recession? And we'll talk a little bit about why we thought -- why we think we had such a roller coaster of a ride in the past couple of years and how we think that might be different, whatever happens next and whenever it happens. So we'll spend our time trying to go a bit deeper and get beneath the surface in ways that are hard to in a 45-minute earnings call, and I think that will be fine.

Awesome. Yes, very excited for that. I think that's going to be really great for a lot of investors. We'll open it up for live questions from the chat, and I see a couple in here, but give you guys another minute to pace your questions in there. While we're waiting, I've got a quick one for you, Sanjay. You ended your opening remarks on the earnings call with excited about 2025 IM. Was that some Yoda-speak? A few Upstarters seem to have thought so, myself included. I wanted to get your thoughts.

You caught that on...

Yes. Nice Easter egg.

Yes, that's sort of an inside joke at the company, to be honest. A couple of weeks ago, we got the whole team together, and we kicked off the year. And at one of the presentations, I did a bit of a Yoda bit. And so that was just maybe a bit of an illusion. I wanted to do it in the Yoda voice, but our awesome Head of IR, Sonya, politely vetoed that. So I just in my -- I wanted to say something excited about 2025.

Awesome.

That sounds like Hermit the frog. So...

Well, it was a cool Easter egg. A lot of people caught it. So it was interesting. Okay. Well, I'll take some from the live chat here. Kaleb asks, when entering mortgage, are you considering ways to create value beyond the loan process, creating value for the homebuyer/seller or realtor? I'm thinking along the lines of what you did with Auto, but maybe to a lesser extent.

It's a good question. I think the short answer is we don't know. We've learned a lot about auto and how dealerships work and how their software stacks work. And so I suspect we always enter a credit segment with the same 2 core value propositions. One is, we want to be better at avoiding that minority of people who will default or not pay you back and therefore, give much lower APRs and higher approval rates to everyone else. Like that's a core value proposition. And of course, we also want to automate the process and make it frictionless for everyone and for anyone who's bought a home recently, how onerous that process can be. And so we want to make it very simple, and we want to make it cheaper and more accessible for everyone. And then I think as we sort of learn about the space, there's always these ancillary opportunities for improvement, whether through the application of clever prediction models, machine models or just more sort of basic digitization and automation. So I don't know what those opportunities are in purchase mortgage yet. To be honest, we have our heads down on home equity lending, and we're very focused there, and we want to make that a big sort of exciting opportunity. But as and when we turn our attention to the broader purchase mortgage market, I expect we will have ideas.

Awesome. We got another question come in here. It says, Dave hinted about international expansion and acquisitions as a potential use of cash. Can you expand a little?

Probably not too much other than, of course, it's an opportunity. I mean, I think we have our hands full in the U.S. credit market, and it is the biggest and most developed one in the world. And in a sense, we've just nibbled on a little corner of it, right, which is unsecured lending is probably the fastest-growing category within credit, but it's also one of the smaller ones. And so there's this constant debate about like how much do we focus on our core business, how much do we focus on the dimension of product expansion within the U.S.? And how much do we think about international expansion? And where I worked prior, we used to have this 80 -- sort of 70-20-10 framework, where 70% of what you're doing is your core, your bread and butter, you want to always make that better; and 20% of what you're doing is sort of near-term bet that for us would be things like HELOC and Auto and small dollar lending. So international expansion is maybe in the 10, right? It's like sort of the further afield bluer sky stuff. I mean, obviously, these types of models would benefit credit in any market. The data structures are different in every market. The regulatory regimes are different in every market. So there's a bit of translation cost. So all that to say, it's an opportunity that we, I think, are excited to buy and we believe to be in our future, but it's probably sitting behind further product expansion in the U.S.

Appreciate the color there. I know we're getting close to time, so I'll wrap this up with one final question. And it's a big picture one, so I think it will be a good one to end on. Upstart's CEO, Dave Girouard, recently said, "When it comes to the fundamental application of AI to credit origination, there is nobody in our class." What do you think is the -- like long-term vision for Upstart? Where do you see the company in, say, 10 years?

Yes. I mean, look, I think my boss said it best. I mean if you sort of believe in AI as a technology, and when we were a younger company, that itself was a challenge. Most people thought it was a bit Hocus Pocus. Folks certainly were outside of Silicon Valley at the time back in 2019, 2020. This wasn't on the radar. But now it's come to the fore. And I think different companies that are using it with different -- or for different applications, I think, have proven that there's some very compelling things that it can do. So if you sort of have a belief in AI as a technology, it should follow -- it's not a large leap to believe then applying it to credit prediction could result in a big improvement to outcomes as we've been talking about. So if that's the case, then I think what we want to look like in 10 years is what we've been talking about, which is we want to apply it to all flavors of credit here in the U.S. and ultimately around the globe. And if we do that successfully or if we're part of a movement that does it successfully, I mean, at its simplest level, like a lot more borrowers should have access to money when they need it at much lower rates through a much easier process. And when I use the adjective much -- when I use the modifier much, I mean like much. Like I think half the market that currently doesn't have -- or half the market that should have access to credit right now is currently locked out. So it's like -- it's a big delta. And I think the vast majority of interest rates that are being paid on unsecured products and on most products are because you're subsidizing defaults that the models led into the system that should not have. And the process is hard. It's like hard to get credit. If you go and buy a car, it's like brain damage to get an auto loan. You buy a house, it's like weeks and mountains of paperwork and like it's just like it doesn't work well. And so it should just be a much more accessible, cheaper, better process for borrowers. And I would assume that we would, at that scale, be an interesting big, reliable source of yield for capital partners that work with us and for banks. And so yes, I think we would like to say that we were the ones that brought these interesting new technologies into the world of credit and made it all better. So I don't know. That's probably just a long way of saying what Dave says much more eloquently, but I think it sort of points to the same direction.

Awesome. Well, thank you, Sanjay, for being here and taking the time again to share your insights and connect with retail investors. We appreciate it. It's been a lot of fun covering Upstart and seeing your progress over the past couple of years and really looking forward to seeing what the team can pull out next. So hopefully, I'll see you at AI Day there in May.

We would love that. And let me just thank you, Henry. I mean it's been very enjoyable to -- first of all, I love doing this because you ask really good questions. And I can't say that that's true of all market participants that we engage with. So you know your stuff. But I think more importantly, I think it's been really nice to -- you've done a great job in sort of becoming a representation of the retail voice. And that's been an unexpected pleasure like we've always thought about that's important and how do we accomplish that. And you sort of stepped up and you've helped make it happen. So we would love to continue to sort of hear what you're thinking and seeing on behalf of all your followers and continue to aggregate questions. And I think it would be great to have that retail presence at our Investor Day. So if that's of interest and it fits with your calendar, we'd love to maybe talk to you about whether we can make that happen. So on behalf of the company, thanks for your faith and thanks for bearing with us through a winding journey.

Yes, absolutely. I appreciate it, and I'll clear my calendar, so you can count on me being there.

First, you got to nail your grades. But after that, we'll find a way to make it work.

All right. Awesome. Well, thank you, Sanjay. I certainly appreciate it.

Okay. Take care.

Thanks.