What if (UPST)

This is speculation, but what if the minor slow down in Q3 could have been partly because they were onboarding a large bank and it took more time and resources than expected? There was the recent rumor about Wells Fargo hiring Upstart. It’s likely that integrating a larger bank could have been more challenging and prevented them from onboarding as many new smaller banks as they might have otherwise. Maybe it’s wishful thinking, but I’m trying to think of other possible explanations - that the company would not have been able to talk about in the earnings report for confidentiality reasons.

Another random thought - I’m not sure if this has already been discussed previously (there’s been a lot of messages lately…). I believe that the current inflationary environment (if it’s not transitory) will help upstart in 2 ways. First, people in general will need to take out more loans because they won’t have as much discretionary funds available. Second, since everything costs more, ALL of the loans will theoretically be larger and larger in the coming years which would be gravy on top of the already expanding organic growth.

I ended up trimming 20% of my shares today after a lot of thinking - but I was very conflicted in doing it. Even after the trim, it is still my single largest stock investment (only Bitcoin is larger). 200+% growth YoY and profitable??? And strong guidance that I believe the CEO said was “conservative”. I don’t have the financial analysis skills of most on this board, so maybe I’m just not understanding things, but wow are we spoiled if that doesn’t cut the mustard.



The reasons the CFO gave for the slowdown, as I heard them on the call, were two:

  1. As Dave alluded to, we are methodically expanding our footprint to serve more of the credit spectrum…This increased breadth of borrower profiles in both directions has driven up the volume of applications at the top of our acquisition funnel by a factor of three over the past year, but it’s also exerted downward pressure on the average platform conversion rate, which at 23% is up 780 basis points versus last year, but down sequentially from 24.4% last quarter. Borrower segments that are relatively newer to our models will initially tend to convert at a lower rate than those segments for which we have longer history. The percentage of loans fully automated on our platform is also down slightly to 67% in Q3 from 71% last quarter for similar reasons. Newer borrower profiles will tend to have more conservative rates of instantaneous approval until we develop a longer history and greater loan volume for our models to train on.

The essence of this is that with tons of new borrowers from different segments, their AI model has to slow a bit to learn something new. That was a slight bottleneck for conversions, but for reasons I found very encouraging.

  1. As a compounding factor, this past quarter we experienced a large coordinated effort to obtain loans fraudulently from our platform. We bore no meaningful financial impact from this activity, but our increasingly public profile as a Company leads us to expect that episodes of this type will become increasingly common. This new reality has motivated us to implement additional protections to our origination processes. And these defensive measures has contributed modestly to lower automation rates.

I’ve seen lots of angst over the appearance of the word “fraud” and you could almost hear the gasp on the call. But the most important piece for me was not that lots of people were suddenly trying to defraud Upstart. That’s not what they were saying. It was a large coordinated effort. One major attack that a single party coordinated. A cyberattack, but more like a DDoS attack on their loan platform than the typical ransomware thing so common in other companies.

We already knew anecdotally that their AI model catches potential fraud very quickly. And they say the attack did not have a meaningful financial impact. But it was a new kind of threat that came with their increasing profile and they had to divert resources to create the additional protections they mention.

So they had a large, coordinated onslaught of fraudulent applications and, from the report, it sounds like they handled it like a boss. Those applications did not get through. Their reputation was not tarnished with their banking partners. Upstart customer’s data was not breached. They had no meaningful financial impact from a large, coordinated attack on their platform. And that was before the extra measures they took after the attack to head off such a thing even earlier.

The Upstart team is exceptional by every measure.

I was disappointed that nobody asked about Wells Fargo. I mean, come on. But I did wonder if there might be a USAA announcement coming.
–They were featured on Leaders in Lending two months ago.
–There were two mentions of the military on the call:

First in Girouard’s opening remarks: But importantly, we’re building a bank-ready product at bank-friendly APRs, always operating within the 36% rate cap prescribed to nationally chartered banks and to those who serve U.S. military service members.

Second at the end:I would like to reiterate Dave’s gratitude to all of the talented teams at Upstart whose hard work makes all of these results a reality. And to also wish everyone a happy upcoming Veterans Day.

On behalf of Upstart, we want to express our gratitude to the women and men who serve our country and to those who have sacrificed to keep us all safe.

I don’t listen to a lot of these things, and it was two days before Veteran’s Day, but did other companies give a shout-out to vets in their earnings calls this week? Maybe it’s nothing, but since I had just been sitting with the list of guests on the podcast, it did make me wonder.

I’m still way overweight at 32% of my portfolio in UPST, but while the hit to the stock has been painful, I still have trouble seeing this ER as a reason to cut any loose. I don’t know when they will hit the tipping point of large-scale adoption, but I still have confidence that they will; and I want to be ready.



From the Yahoo UPST message board: "Upstart founder Paul Gu makes a very apt observation about why big banks will not build their own AI platform to ascertain credit risk. And if they do, it won’t be nearly as accurate as Upstart. The reason is Upstart’s AI tracks loan repayment factors across all their lending partners. So the information Upstart obtains for their AI platform gets really smart really fast. And will only get smarter and smarter over time. Lets say JP Morgan decides to start an AI credit risk division. JP won’t have the volume of repayment events to plug into their system to scale it out. All the while Upstart will keep getting further and further ahead by evaluating repayment events across all the banks in the Upstart network. And it’s not as if JP Morgan or any other legacy bank could even come close to Upstart AI efficacy. "

UPST’s AI is “learning” from the data collected from all those partnered banks while any JP will only have that one JP dataset. AI is dependent on data, the more the better. When viewed from this perspective, UPST has such a huge first mover advantage in AI. I am sorry I trimmed my UPST yesterday.