UPST: post-earnings thoughts

UPST (21.4% allocation; Q2FY21 10Aug earnings date)

Coincidentally, I’ve saved the best for last. After the seeing the massive 20% raise to annual revenue guidance (after Q1), I was expecting great things for Q2. The results for Q2 were spectacular. In fact, I’d say that the business growth was significantly better than ZM’s Q2FY21 blowout result (+355% revenue growth). UPST delivered 1018% revenue growth and 60% sequential revenue growth! Yes, comparing revenue to last year’s Q2 is a bit unfair so let’s look back to Q2 2019 and annualize the growth: we get 496% growth (Q2 2021 over Q2 2019) which annualizes to 223% CAGR over the two-year period. It also looks like growth is accelerating, and, to top it off, there are headwinds for lenders in the unsecured personal loan market! These headwinds are present because government financial support acts to subdue the demand for loans. A large portion of the pandemic related government support will drop off in September so this headwind will soon abate. Unlike ZM’s quantum leap growth in 2020, UPST’s growth seems durable, and, as an investor, I’m expecting hyper growth to continue for a while. The 25% increase to annual guidance (after Q2) is a very strong clue that Q3 is going to be another great quarter. The stock price exploded after the result to close the week up 50%. I believe that, given the business result and future potential, the share price increase is justified. When someone asks me whether I’m going to trim my position, I say, “No way Jose!”. I’m clutching on to every single share that I own!

What’s so special about UPST’s business? It all hinges on UPST’s ability to predict who’s going to pay back a loan. If UPST can more accurately distinguish which prospective borrows will pay back and which will default then UPST can increase the profit for a lender. In addition or alternatively, it also enables lenders to gain marketshare by offering more competitive interest rates to its borrowers. And then you have an instant underwriting process (i.e. Upstart’s AI makes an instant decision); by comparison, banks traditionally use a team of underwriters (i.e. paid employees) who can take hours of their time and days before they approve a loan for disbursement. Thus, the lender isn’t just getting a lower default rate; it’s also getting faster and lower cost loan approval. In exchange, UPST gets a cut in fees for each loan that’s issued. Great deal for the lender, great deal for the borrower, and great deal for UPST. As long as this all continues to work, it’s also going to be a great deal for us shareholders.

So is it going to keep working? So far it has been working. Some folks on Saul’s board, in particular jonwayne235, have analyzed and compared the performance of loans issued via UPST underwriting AI and other lenders. In aggregate, UPST’s loans have been winning hands down every time. jonwayne235 even applied for an unsecured personal loan at various lenders. The result: UPST provided the lowest rate and did so instantly. Now, jonwayne235 is not just very smart and analytical, but he’s also financially responsible because he claimed to have a FICO score over 800. How would UPST do for people with lower credit scores? Well, jonwayne235 told me that another guy, who has a much lower FICO score, also received the best rate from UPST. Someone with a low credit score would have a much harder time getting a loan. Such a person would also need to pay a much higher interest rate on the loan. Thus, if UPST can identify people with poor credit scores who will pay back the loan, then that borrower will be extremely profitable for the lender. So far, UPST’s AI lending decision algorithm seems to be working. It’s further evidenced by an increasing number of lenders (now 25 stated by CEO Girouard during the Q2 earnings call) reported to partner with UPST. One Upstart partner-lender is now dropping the minimum FICO score requirement from its lending decision. That’s incredible to me, and it shows that UPST’s partner-lenders are beginning to place increasing trust in UPST’s AI; this may bode well for revenue growth as increased trust in the UPST AI should lead to increased utilization by the lender partners.

Most of UPST’s loans end up getting sold to institutional investors. UPST claims that it has partnered with 150 institutional investors to buy up the loans that get originated by UPST’s 25 bank and credit union partners. On the earnings call, management was asked what’s limiting the number of loans that get funded using UPST. The answer was it’s not demand from the institutional investors that’s limiting growth. The limitation is on the origination side so currently growth is limited (if you can call 60% sequential revenue growth limited) by the number of willing borrowers UPST can run through its AI and pass to lenders for funding. Thus, future growth will be fueled by increasing the number of bank/credit union partners and increasing the utilization of UPST by those partners. Progress in signing up new lending partners has been steady and has almost double since I started following UPST (February 2021). I suspect that the AI’s success is catching on in the banking industry and another lenders may soon jump on. If that does happen then we could be in store for an unexpectedly explosive disruption event (pure speculation on my part). In addition, UPST is launching a Spanish version to target the large and sometime underserved Latino population in the U.S. This effort could boost growth further especially since competition is much more limited for lending to this demographic.

UPST’s next target market is the auto lending market. UPST has been working hard to plant the seeds for 2023 success in this market. The auto lending market is 6x as big as the unsecured personal loan market. In the last quarter, UPST expanded into 47 states (from 14 states at the end of Q1) and has funded 2000 loans in 40 states. The first five lenders signed up to offer auto loans (via UPST) during Q2. The auto lending market could get big for UPST in 2023. After that, I expect UPST to continue to disrupt additional lending segments.

Yes, there are risks with UPST as an investment, but, for me, the risk-reward flipped very strongly to the reward. The lending market is enormous and it already exists. UPST’s process has very little friction to growth and switching the underwriting process to UPST is also frictionless. Therefore, I could see growth exploding unexpectedly at any time. Think about that for a minute: if UPST suddenly gets a ton of more business (it’s possible because of the low friction that I mentioned) then as investors we could be suddenly surprised by a result that moves the stock up fast. The rate limiting factor remains signing up new partners and getting the partners to ramp up their use of the UPST AI. I like the risk-reward so much that I’ve made UPST my largest position.


UPST delivered 1018% revenue growth and 60% sequential revenue growth! Yes, comparing revenue to last year’s Q2 is a bit unfair so let’s look back to Q2 2019 and annualize the growth: we get 496% growth (Q2 2021 over Q2 2019) which annualizes to 223% CAGR over the two-year period.

There’s a calculation error above. The 2-year CAGR is 144%, not 223%. The mistake was found by Saul which doesn’t surprise me at all. There’s a reason he was the captain of his math team in high school!