All right.
All righty. Welcome back. I am very happy to welcome Dave Girouard, founder and CEO of Upstart, to our conference today. Dave, thanks so much for joining us. It's always a pleasure to talk to you.
Thank you.
That's a fresh glass of water for you as well. Before we begin, I'd like to quickly read the following. Today's discussion may contain forward-looking statements that relate to future results and events, which are based on Upstart's information available as of today and are subject to risks and uncertainties. Actual results may differ materially from these forward-looking statements. The discussion may also include non-GAAP financial measures, which are not a substitute for GAAP results. Please refer to the company's filings with the SEC and its IR website for additional information, including GAAP to non-GAAP reconciliations, along with other disclosures.
Good stuff.
Yeah, good. I think we're done now. No, I'm just kidding. I attended your AI Analyst Day last week, which was super fascinating, and I want to delve into business. Maybe first you could give us a little macro commentary, a little macro flavor. You guys have a privileged view of things. What are you seeing out there in terms of, you know, state of the consumer?
Sure. Yeah, our business is very attuned to, consumer financial health is probably a way to put it. A little different than what most of the market pays attention to, which is maybe just like how much they're spending and if they're spending, you know, we don't have the same interest that, say, a retailer does who just wants consumers to spend and spend. Our view really is, is more how are they, how are they in terms of, are they saving, are they healthy, are they, you know, obviously making good on their, on their credit?
Generally speaking, I would say consumers are very steady state these days. They deteriorated pretty dramatically when stimulus was withdrawn way back toward the end of COVID. Though they have improved since then, they're not, they're not sort of back to what we would consider to be a normal long-run state. We have this way of measuring it. We call the Upstart Macro Index, and it has improved, but it's still what we would consider to be elevated, meaning there's significant risk priced into our loans. That just puts us in a place where we are, we're not concerned in any way.
We think there's probably more upside than downside in the state of the consumer. What we would like to see is, you know, consumers are returning to the long-run sort of personal savings rate. You know, if you looked at the personal savings rate, which is in our mind, it's a Fed number, if you looked at the long run, it's typically maybe 8%, sometimes 9% range. It has been very low for a long time now, more like 4%, even down at sometimes 2%-3%, which sort of suggests, you know, the finances for the average American family have been a bit upside down with inflation and all that's gone on through COVID and stimulus, et cetera.
I think it's a steady state. It's not a worsening state. It's really, obviously important that inflation has dropped pretty dramatically. That's helpful. Hopefully tariffs won't upset that too much. We have a, you know, and when we give guidance or anything, we don't assume any improvement. We don't assume rates are going down. We don't assume the macro changes in any significant way. We feel, I would say, pretty comfortable where it sits right now.
Good. This, this being the topic of your, your event last, last week, AI, you guys were sort of first to the dance there a little bit before it became more of a trendy, buzzword. A lot of other lenders also talk about deploying AI. You know, what is Upstart's sort of sustainable advantage when it comes to AI? How do you differ from, from the pack, as it were?
Yeah, I mean, I would just say that since the fall of 2022 when ChatGPT came out and the market sort of swerved so much toward AI, I guess you'd be sort of irresponsible if you somehow didn't say that you do this and weren't doing something. Certainly it's become a thing where businesses I can't even imagine have anything to do with AI suddenly are AI this and AI that, which is fine. It's the natural reaction of the shiny new thing.
The core of our business since the day we started has been about better risk models through very sophisticated math and lots of data, which of course is today known as AI, to make for a far better credit product. As much as other people are saying it, what we tried to do in our AI day last week is show really the depth of what we're doing and how it leads to a pretty radically better consumer product. One fundamental way to think about it is in a lot of places, I mean, J.P. Morgan Chase has, what do they say?
How many engineers they have, how many machine learning or what have you. There's a difference between people sitting in a corner of a research shop building models and doing things on themselves and having models that have a daily, weekly train into production where this is getting X% better every single week. That's what real AI is. It's not about research. It's not about like, oh, I discovered this cool new thing and now we're going to go to the committee and review it and see if we could implement it.
You know, AI at scale is a little bit like you see from OpenAI or from Gemini, from Google. It is a constant move toward more accurate, better models that are in a production pipeline, right? Literally from machine learning engineers, testing, trying, validating to into production in a matter of a few weeks. That is not something we see any evidence of happening elsewhere among our peers.
There's certainly parts of FinTech in payments. There's always been a lot of very sophisticated models to identify fraud. That's certainly been out there for some time. I think there's probably like, very short-term stuff like BNPL has all sorts of very interesting risk models. I'm quite sure they're doing, but in our class of loans, which are more longer-term loans of different types, we don't think there's anybody that are building models similar to ours.
You know, I think when people think of AI in the context of your business that, and you mentioned this, they sort of zero in on the ability to underwrite. I think one of the underappreciated parts of your story is how you've leveraged technology to improve the customer experience, speed up the experience. You know, maybe you can, you know, walk us through sort of that idea and also maybe sort of a, I thought this was a creative idea, sort of a before and after for a loan application, maybe what folks would've faced versus what they face now with you guys.
Yeah. I mean, one thing I think is really important, a loan is a simple idea and it can be a commercial loan, real estate loan, personal loan, mortgage, boat loan, doesn't, doesn't really matter. The thing about any type of loan generally, there's only two things that matter, the price and the process. Everything else is just sort of noise. We have intense focus on improving both on the process, which is a little bit what you're referring.
Nobody really wants process. They just want a loan. There's a myriad of things the industry, depending on the type of loan, has put in front of somebody, mostly because of this necessary risk mitigation, right? Very long applications. Maybe they call your boss. Maybe you have to go in, in person to do closing. You have to submit a lot of documentation. You probably have a phone caller. There is just a myriad of things that are necessary in the traditional sense of lending to de-risk the process.
What we have been able to do over time is just build more sophisticated models to verify things in the background very, very quickly and use models to make sure we can predict things that we are not able to verify in a way that today, 92% of our loans last quarter, there was no process. It was instantly approved. That is, you know, the ultimate process is no process. If you went back to, you know, say beginning of 2017, when we really started down this path, every borrower had a phone call with a credit analyst.
They uploaded four plus documents per person, you know, a driver's license, a pay stub, maybe a tax document, proof of residency, something. Every time you uploaded a document or asked that a document be uploaded, you would lose 15% of the borrowers. You can imagine like the conversion hit to saying, oh, sorry, can you now get your W-2 from last year and blah, blah, blah? Like it's just, yeah, no, thank you. What we've just done over time is eliminated all that. The difference is radical.
The conversion, when you instantly approve somebody in the moment, is about 3x what it would be if you sent them off on a journey to get more documents. That experience, you know, we think of the world as like roughly half people care about getting the lowest price and they'll go through hell and high water to get the lowest price. The other half just want it to be easy and they'll pay a little bit more if it can just be easy. We push hard in those two dimensions to just make the product better and better.
One notable detail I thought from the recent earnings call was that super prime customers now account for, I think, a third of loan originations, personal loan originations. Talk a little bit about that, that detail and that potential shift, you know, and maybe give us some color about whether that's happening across multiple loan products or one product?
Yeah. You know, we grew up really with the differentiated risk models that could identify what you might think of as hidden prime, people that were not obviously prime to the rest of the world, lower credit scores, no credit scores, recent immigrants. That is kind of where we differentiate ourselves as we began, and still it is very much, we are good for a lot of reasons, w e decided that was not the ultimate form of the company and who we wanted to be. Part of it is just an interest in diversity of serving not just part of the population, but serving everybody.
Part of it was really just a simplification of message. The ultimate expression of AI in this market is you should have for any particular person the best rate they could possibly get and no process. You can't do that if you're only serving half the country, right? We decided about a year ago that we're going to aim to have the best rate and the best process for everybody, for every product we have. What we showed in AI day is we've actually achieved that. About a third in the personal product, about a third of the borrowers are super prime.
In each category, super prime, mid prime, or non prime, we have the highest win rate of any platform out there. That is really the end result of better AI and building the business. I think we're headed toward a world where, whoever you are, we think Upstart should be in whatever credit product you might seek. We should be the first stop and maybe the only stop because it'll just be there. It'll be the best rate guaranteed, and then you'll have to do nothing else to get that loan of whatever type you need. There's nobody else. That's a category of one company if we can achieve that. That's exactly where we're headed.
When it comes to feeding applicants into the top of the funnel, you know, has your distribution mix changed over time? What are your largest channels now, maybe relative to, you know, the IPO time?
The thing that is probably the largest channel that was much, much smaller then really is our own customers coming back. You know, when we were totally new, the vast, vast majority of loans were the, the applicant did not know us, had never seen us before, and then found us and got the product. Now, of course, we are, we've been around a lot longer. I would say 30%-40% of the loans are someone who's already gotten a prior product from us.
That has become our own sort of user base, is our largest source, which is great because you don't have to pay to acquire them. You're not, but beyond that, it hasn't changed that much. You know, aggregators are useful and important for first-time borrowers, people like Credit Karma or LendingTree or people like that. We do a lot digitally. I will just say generally, given where we've gotten to in the last couple of quarters, meaning this notion that we believe we have the best rate for everybody, or we certainly like statistically have that, you're going to see just us a lot more being direct to the consumer and having a proposition and a brand that will resonate.
Not to say aggregators can't be, can't be good for us. You know, I think that idea that we have a simple, very unique proposition to the consumer will begin to resonate more and more. Right now, honestly, the product is ahead of the brand. You know, if you just read about what someone will say, it was like, it's the best loan for someone with a bad credit score, something like that. We have a lot of work to do, but I think it'll become to this place where, you know, as we've called it, the always on, everything credit store. That notion, I think it has a lot of potential to it.
Another concept you introduced at your event last week, obviously at your event, left a big impression on me. Another concept you introduced last week was that your models, product mix, a few variables today would kind of make the business, you know, potentially prove more resilient to macro stress. That, again, was the case a few years ago. Maybe you could kind of revisit that idea for us and flesh out a little bit?
Yeah, for sure. I mean, one of our goals is just to make Upstart a resilient company that will go up and to the right and grow and grow profitably. I don't think you can have zero macro exposure. Every, I mean, I was at Google for eight years. Google had pretty significant macro exposure. There's like, not like every business doesn't have some, but I think there is an 80/20 rule where we can get rid of 80% of the macro volatility and we're on a very fast path to do that.
You do that by having a diverse set of borrowers. That's part of what super prime is about, which is, you know, they are affected differently or at different times than a less prime borrower. Having different products, secured products tend to be less affected. Having a home equity product and having an auto product I think is really important. We are going to move toward other types of business models that I think will drive us toward a much lower volatility state.
Generally speaking, to build something so unique in our view, we are again the only company building such a thing with such a proposition to the consumer, is I think in some sense, the opportunity to build something this large and this unique may come with a little volatility. I would just say, look, if we're going to be a company, you know, sometime in the future with $10 billion in revenue and $2.5 billion in EBITDA, would you care exactly what the path is to get there? I think the uniqueness of what we're building is so strong. We are moving to less volatility.
The other maybe obvious thing I should have said is, today two- thirds of our funding are in long-term, you know, committed partnerships as opposed to at-will funding. In the last few years, we have really completely revamped the business to head toward a more predictable growth path, through cycles. We may be in one right now, right? We've had no pullbacks in funding from either banks or credit unions or in our private credit partnerships. We are on a pretty dramatically stronger footing than we were a few years ago.
What makes you guys more attractive to these capital partners, you know, or what makes you not more attractive? What makes you attractive to these capital partners? What are they looking for in you? What are they finding in you?
I mean, they have capital to put to work in private credit, just like private equity, its cousin or sibling, you know, has long-term yield targets that they want to be able to put money to work for a very long period of time. We have a great product for that. They can get it in a very ratable way, meaning monthly, and, you know, with the proper structuring, can get exactly to the targets that they want. You know, the credit, credit's always going to move a little bit. There's no perfect model. It's never going to be like, you know, not going to move whatsoever.
The alignment is really good. We co-invest with them, so we have skin in the game and we're interested in their support for a long period of time. They're interested in us in delivering reliable yield. It works really well both ways. I think it's a structure that evolved as part of the grander private credit boom that's happening. That boom, by the way, I think is an incredible benefit for so many industries.
Banks are regulated in such a way that they are risk intolerant and they have pulled back on risk, or in lending generally. They've lost market share in lending for a very long time. Private credit, I think, is a great structure to sort of supplement and complement what banks have been doing forever.
You mentioned co-investment. That's something I get questions on a fair amount. Talk about that shared risk. Dimensionalize that for us. Help us understand to what degree is the risk shared. Any color there would be helpful. Yeah.
Yeah. You can think of it as these have evolved a bit, right? The first of these agreements we did a couple of years ago and they were entirely bespoke and new and they've sort of, as we've matured, they've gotten a little different and better. Generally, you can think of them today as a joint venture in an entity that is created where we have some equity capital in it and they have the much, much larger portion of the equity capital in it. Then there's some senior financing in it.
That entity becomes a loan buyer. What that essentially means is we have some credit exposure to what we do along with this adjustment, but it's quite small. If you think of our businesses, you know, our take rates, they can vary, let's just say 8%-10% take rates. And we have a, you know, a much lower single-digit sort of fraction that we have exposure in. Also, really importantly is we have control over all the dials to make sure that this risk doesn't become a problem for us in the future.
It's also structured in a way that we can contribute on good months or good quarters toward this thing and then have room to have less good quarters. That's the sort of point of it. It's structured to be long-term. The thing that we have shown very reliably since we started is if you look over the long haul, the yield on our product is exceptionally good. These are structured to work over a long period of time, in our interest and our interest in, and I, I, so we're excited about it.
I, I do think, by the way, we know our technology better than anybody. We control the dials better than anybody. We should have risk exposure. We should be a party to this. I think building the platform that we're building, it's the right, it's the right structure.
I think that's a critical insight that you control the dials so you can have some control over what is the end size of that, of that exposure. I want to get into some of the new products that are not so new anymore, but some of the products over the last couple of years that you've kind of been getting into. First, maybe on sticking with this capital partners theme with auto and home equity, what is the capital partner backdrop there? Do you need new partners or are your existing partners interested in underwriting those loans as well? What,
I think it can be a combination for sure. The banks and credit unions love both those products, auto and home. In fact, frankly, they like them more than they like unsecured loans. There is enormous appetite in the lender part of what we do. I think it will bring in lenders that would not otherwise work with us because they have less interest in unsecured lending.
Home lending and auto lending is tried and true in that world. That is true. On the private credit partnerships, I think generally we definitely expect and we are already working on multiple newer products going into those agreements. There is just no reason you cannot blend them in. It is actually better to have a more diverse set of products in these relationships. For us, it is actually enormously valuable.
It's kind of like a defragmentation of our funding supply, right? Suddenly we don't have to go, oh, this is a HELOC. We have to go out to all the resi buyers out there and structure new agreements. Actually, no, it's just another form of credit that has a certain attach to an unsecured loan. I think we're very optimistic that these partnerships will expand to include multiple forms of credit, maybe not every one of them, but I think quite a few of them will.
That's to everyone's best interest. By the way, we are in the process now for both home and for auto of kind of moving from things that we're growing up on our balance sheet toward moving them off balance sheet toward these types of partnerships.
Give us a status report on auto. That's an area you entered a little while ago. You know, what's going on there? What have you learned since the initial rollout?
Yeah, two separate products. You know, right when we, you know, the first thing we built was an auto refi product. Then through an acquisition, we started getting into auto retail, the idea that you can get an Upstart-powered loan at the place you're buying your car. They both, for a lot of different reasons, have taken a long time to go, to be frank. We've learned a lot. These were the first non-personal loan products. It was our first effort to move beyond what we started on.
At a very difficult time, anyone who knows kind of the auto industry since COVID and post-COVID has just been a circus of there's no cars to sell at all. There's cars, but they're too expensive. Interest rates are going up. A myriad of challenges has made it difficult for us. We definitely have made some mistakes along the way. Having said that, they are really starting to come together nicely now. Credit is performing. Conversion rates are improving. Distribution's improving. What you saw is in the auto business, it grew, I think, 42% quarter-on-quarter. The credit performance is very good right now. It has taken us a while.
To be honest, our patience was not unlimited on this product area, but the effort to stick with it and know that fundamentally this is a product area that makes sense with a lot of inefficiency, a lot of displeasing processes, and we're really getting there. I'm very optimistic on it. It has been a long road. It could have been a time where we would just say, look, this is enough at this, we should stick with what we know. Truth be told, it's really coming together finally, really well. I'm excited about it.
HELOC has been much more straightforward. It's worked well from the beginning. It's just ramping and building. We've never had, I mean, any misgivings about it or anything. Thankfully though, I would just say one, you know, you have different children that have different behavioral characteristics. You know, they're definitely two different children. Fortunately, they're both now at a place that we're excited and moving in a great direction. You know what?
I mean, I always said we also have a small dollar product, which is just, again, always done exceptionally well. It's one of the easy children, if you will. I always thought of these like four new products that we have. If we could get two or maybe three of them to work and the other ones ended up not working, that's what startups do. That's what, you know, innovative businesses do. I think right now it's more likely that four out of four are going to work. That is to me incredible success.
It sounds like the value proposition with the auto and home products has a lot to do with the experience. They're, I'm guessing maybe a little easier to underwrite because there's collateral, maybe you can tell us, but it's really changed the experience with these products. Is that a fair, you know, statement? How have you changed the experience for these?
Yeah. I mean, secure products, of course. The problem, I always call the auto refinance business a zero billion dollar industry, right? Because most people don't know that you could refinance your car loan. They assume like, I got this loan, it was the best I could get at the dealer. And, and so why do I think a year later that suddenly I'm going to get a much better loan? And also if I did, God knows what that would involve, right? I'm probably going to the DMV. I'm probably doing X, Y. It's just, that's why it's a zero billion dollar industry. We've built a product where finally we can say this will happen really quickly, really easily.
And by the way, we don't go out and try to like educate the whole world about that. It's too, it's too burdensome. We just cross-sell it to people that already know us. The example we kind of showed is this woman came in, got a small loan to pay off some debt. We were like, we see you have a $400 a month auto loan and we believe we can lower the price of that. She was, and literally, you know, in a matter of 45 minutes, she got her first loan. She was happy with that. She saw this thing and said, wow. She reduced the price of her monthly payment on her auto loan.
That is the kind of thing where you just have to have a better product. You can't take weeks to do it. You can't send people to DMV. That is what technology can do. This is what we've done in the unsecured product. It is definitely a bigger hurdle in the secured products. But guess what? That means there's fewer people doing it. It's actually easier from our perspective to create a differentiated experience. In personal loans, as we said, 92% instantly approve. No, no problem. I don't see any reason we can't over a matter of years push the secured products in the same direction.
As high, I don't know, but we are just getting to the point where we can do the first fully automated auto refinance loan. And we'll do it very soon with the HELOC. And that's a totally new world. That's why what we described in our investor day was that imagine a future where every credit product you might want is available to you at any moment in time and there's no process and is guaranteed to be the best rate. That's a different world than what we live in today.
That's the kind of thing you should expect from AI, right? Like this crazy leap forward where you never go somewhere to get a loan. You never even have to fill in any information. You can be assured that it is the best rate you will find anywhere. That's a future that we are pretty rapidly moving toward. If we do that, I don't know who else is doing it. You know, that's why we feel good about the proposition.
That's fantastic. I forget which business sage said that the purpose of a business is creating a customer. That sounds like it's exactly what you're doing in this, in this context. You touched on some other recently, some other potential products, including maybe revolving subscription models and something that I thought was very cleverly called servicing as a service. Maybe share with us what those ideas are.
I mean, first of all, we do believe in the notion of, and this is a little borrowed from Amazon maybe, but the everything store for credit. Like, you can't just have one or two products. If you want to be the place that everybody wants to be a member, wants to have this app on their phone, then you kind of need to cover all bases or it just isn't particularly interesting. You should definitely expect more products from us, a revolving product, more home products, you know, who knows what else, but you should expect a full roster of products that are the best at what they do in terms of price and process. That's wholesale. What was the other part of the question, Ram?
It was just, a couple of products that you mentioned again at the event last week.
Oh.
subscriptions.
Subscriptions, revolving, and then servicing as a service.
Yeah.
Subscriptions, of course, like there can be a place in time where we could end up having a service that the consumer subscribes, pays subscriptions for, and just has an elevated service of lots of types. That's possible. It's not, so you can just think of that as a different way to monetize the types of services that we would offer. Servicing as a service is a little different. It basically says if we get to a place where we're extremely good and differentiated at servicing loans, there's no reason we couldn't take that out and service other people's loans.
From a business perspective, that would suddenly be a very nice countercyclical thing to originating credit. There's just, you know, I think if we are true to what we say and AI is going to build very differentiated products and we don't see others in this very conservative market building such things, I think the chances to monetize and take this business in many ways are going to be there for a long time. The sweet spot of it really will be, and I think one of the things I feel happy about just in even most recent months is really clarity on who we want to be and what our value prop is and why it should matter to the world and all these things.
It really does come down to just a superior consumer products, and that, you know, selects for the best borrowers. It selects, it makes the capital want to be part of you, be part of your story. That is what we're building. It is a, you know, it is a credit unbundled story. You know, everyone else is, I'm a NeoBank or I'm a super app and credit's just one of the 10 things I do for the consumer. Just like if you went way back to the first dot-com, you know, Google was like, I'm not building a portal. I'm building search.
Everyone else thought search was part of the portal. Now, if you're younger than 40, you probably don't know what I'm talking about. We're a belief that credit is very hard to do, really hard to do well. Consumers will do whatever to get the right product. That means a dedicated, credit destination with the very best prices, best process is going to be a very, very large proposition.
Fantastic. I think we're just about out of time. Great conversation. Thank you so much for being here. I appreciate it.
Thank you.
Thank you.
My pleasure.
Thanks all.