Upst Business Evaluation

Hi All,

I invested in Upstart during two occasions with a modified cost basis of $90ish after the May 2021 selloff, by which, it was my largest position. After seeing the constant headline quarterly earnings and upbeat management guidance revisions, along with a surging stock, I became complacent. Rather than monetize the investment after a >400% increase in return on my investment, I felt the long growth runaway, new growth markets, and huge TAM management was citing were sufficient to hold this stock for a very long time. Unfortunately, I sold this investment this week at only a 10% gain after several weeks of declines made me dig deeper and reevaluate my thesis. Below I will highlight some of my new interpretations as I did further due diligence, and hope it may aid any current or future Upstart investors. As of now, I’m sitting on the sidelines as more interesting opportunities have come up in the recent selloff.

For all the beginner investors and also seasoned ones, selling is the hardest part in investing besides portfolio positioning/risk management. Everyone has their own strategy, but trying to pick the top or bottom is incredibly hard, and no one ever went broke taking a profit. I’m sure Tesla’s parabolic surge along with some of the meme stock mania affected my psyche with regards to selling Upstart once it reached ~$400 given this was a far superior stock/business in my view than meme mania stocks. In addition, I wanted to reduce my tax bill by holding until May of this year so I can qualify for long term gains on the stock. Unfortunately, the timing of the stock collapse and market conditions were just an unlucky combo of events and this was a valuable lesson for me regarding creating a future framework when it comes to selling. Now onto some due diligence for Upstart. I think everyone needs a better understanding of what to consider besides what’s already been mentioned so they can value it properly (disclosure: you will need to value it yourself!). Here we go!

1.) Upstart is an AI lending platform that has three sources of revenue streams: (a) referral fee (referring an applicant to a bank) (b) platform fee (fee of originating loan on a bank’s behalf); and (c) servicing fee.

2.) ~70% of loan applications are automated. Primary leads are Credit Karma and

Benefit: By automating 70% of the loan applications, Upstart has a very unique proposition to banks. Remember, Upstart is not trying to become a bank. It’s in the business of helping and working with banks so that they can leverage their balance sheet and make better and more risk adjusted loans according to Upstart. If you can eliminate the majority of loan officers in a bank, that is a major cost savings Upstart is providing. Additionally, for banks that are capped in terms of loan capacity due to not having enough personnel or not finding it suitable to allocate resources to hiring more loan officers or whatever the reason may be, the smaller banks and credit unions can add incremental revenue without adding more fixed costs using Upstart’s AI lending platform.

Risks: Intuit purchased Credit Karma and there is risk for contract renewal with a new parent (maybe Intuit takes it’s data on taxpayers and couples it with a lending solution of it’s own). No one knows what Intuit will do, but there is some comfort that Upstart’s business is hugely beneficial to Credit Karma. However, Upstart does also generate revenue from it’s own websites and spends millions of dollars sourcing customers via mail & ads.

3.) Bank partners retain 20% of the loans originated using Upstart and the rest is sold and repackaged as ABS via firms such as Cross River Bank. I’m not sure if this is good or bad, but maybe if banks were holding a higher percentage it would be a greater vote of confidence. Perhaps everyone is trying to make the “easy” money and watching from the sidelines as they monitor the loan performance to get more comfortable in the future.

Benefit: Provides banks with liquidity and can make more loans as a result.

Risks: ABS market seizes up or institutional demand for Upstart securitizations decline, causing Upstart to purchase the loans from banks and retain them on their balance sheet.

4.) This ties into point #1. Unsecured consumer lending is vastly easier to enter & more profitable than auto & mortgage lending. The question is why? The reason is the latter two are secured loans and hence the potential fee/take rate by Upstart will have to be substantially lower than the unsecured personal loan market. Thus, it will be more of a volumes game, but it will not be easy for Upstart, especially in the auto lending universe. In auto lending, Upstart has two potential target markets, new and used. Where I expect them to make headway is with private dealers no specific brand or companies like Carmax. Also, the large automakers have their own finance arms and that will be very difficult for Upstart to penetrate. It will either need to spend heavily on marketing to acquire customers or make headway via it’s auto lending software (formerly known as Prodigy) into dealerships by which customer acquisition costs would be the lowest if using the latter. Also, I have no idea how restrictive the dynamic of large auto finance arms and dealerships would be on Upstart’s ability to penetrate major brand dealer chains.

5.) The most perplexing thing to me about Upstart’s revenue growth has been why did they dramatically go bonkers especially given all the stimulus provided by the government and consequently high savings rate. It turns out Upstart made a major pivot when it went public with respect to its model that allowed it to generate more loans. I think the 2021 vintage years are the ones to examine closely but I would not over scrutinize if delinquency is higher than older vintage years. Reason being is Upstart is largely generating non prime loans and it may be wanting to “train” it’s model so that it can get smarter, at the expense of some higher default rates. This should not be a huge alarm in my opinion but obviously people would question their AI advantage and this does need to be monitored over the years.

In 3Q21 Upstart did “fee” revenue of $210.4 from transaction volume of $3.13 billion, roughly implying a fee rate of 6.77%. There is no way Upstart will get this same rate with prime borrowers, or in the auto & mortgage industries. Thus, I think Upstart’s upside is more limited than what management is making it out to be by providing huge TAM figure’s.

Now, with regards to valuation and the mistake I originally made. Upstart is definitely not a Saas company, but it does have some Saas like elements (high margins) and there is great optionality. Also, there is zero possibility that Upstart will get the same take rate in the auto and mortgage industries, thus just going off their TAM is hugely misleading. It is possible down the road Upstart begins selling it’s AI as a software while simultaneously receiving a take rate of each loan generated along with other potential fees. This would definitely make the business and hence the stock, a lot more interesting but as of now, that is not the case. Thus, the conclusion is for the time being, Upstart can keep taking market share in the personal unsecured market, but don’t think the fee rates are going to be anywhere close in the auto or mortgage markets. Thus, Upstart’s TAM isn’t as big as it states but they do have ways of transforming their business potentially in the future via a Saas offering which would make this way more compelling.

Happy investing to all!


I forgot to include this analysis and also wish I would’ve done this sooner myself. Here is a breakdown of Upstart’s potential market opportunity, subject to the below assumptions:

Based on their recent earnings, as I mentioned Upstart had “fee” revenue of $210.4mm based on a transaction volume of $3.13 bln in 21Q3, implying 6.7% take rate. Upstart projects the personal loan market as being $81 bln of which they currently have a market share of 15.46% if annualizing their last quarter transaction volume. Based on forward estimates, it’s fair to assume Upstart will generate more revenue from the personal loan market and continue gobbling up share. How high Upstart’s market share goes remains to be seen but this is by far their most profitable market.

However, as I also mentioned before, there is no way Upstart’s take rate of 6.7% would apply to the auto and mortgage industries. Just for hypothetical purposes, let’s assume the take rate would be 1% for each of these markets and that Upstart’s market share can get to 15% like in the personal loan market. Based on the TAM’s provided by Upstart, their potential run rate revenue would be as follows:

Annual auto lending fee revenue potential = $672 bln x .15 (market share) x .01 (take rate) = $1.0 bln

Annual mortgage lending fee revenue potential = $4.5 trln x .15 (market share) x .01 (take rate) = $6.75 bln

Based upon the above, the biggest variables driving Upstart’s future revenue streams from the auto and mortgage markets would be their market share and take rate. Anyone can fiddle around with these assumptions, but it’s a good exercise to at least gauge what the “potential” could be if this is even considered a baseline. There’s still a lot of optionality as they have discussed plans to enter business loans and payday loans at some point. Their AI should also get smarter with time and as it processes more loans.


Technically Upstart can achieve a crazy high take rate in the auto lending business if they target Credit Acceptance’s target segment and steal market share from them along with from others. I apologize for forgetting about this in my earlier post.