Read on Twitter
Upstart: AI Meets Credit
$UPST is an online lending marketplace that uses AI to make the lending process more accurate and efficient for both lenders and consumers. The business is built on the premise that traditional banks are inconvenient and fail to serve borrowers.
$UPST is an online lending marketplace that uses AI to make the lending process more accurate and efficient for both lenders and consumers. The business is built on the premise that traditional banks are inconvenient and fail to serve borrowers.
The heart of $UPST’s business is its AI lending model that uses over 1600 data point to assess lending risk and determine loan terms. The model has been trained over the past 8+ years with millions of repayment events which have continued to improve its accuracy.
$UPST is disrupting the traditional lending process as many banks continue to use dated models to assess lending risk. FICO credit scores (developed in 1989) continue to be a key part of risk assessments and traditional risk models incorporate at most a few dozen variables.
As a result, borrowers are 1) denied loans that they would repay or 2) offered loans with interest rates higher than they should be. 4 out of 5 Americans have not defaulted, yet less than 50% qualify for prime credit.
With its AI model, $UPST approves 2.7x the loans at the same loss rate. Consumers receive higher approval rates, lower interest rates, and more efficient digital experience. Banks, in turn, get access to new borrowers, lower fraud and loss rates, and increased automation.
$UPST is not a bank and has no plans to be one. Instead, it operates a website through which consumers can apply for loans that are then routed to $UPST’s partner banks. $UPST’s partner banks can also use $UPST’s lending algorithm to originate loans through their websites.
$UPST primarily generates revenue from fees with its partner banks. It charges referral fees ($200-300 per loan it originates), platform fees ($400-500 per origination), and loan servicing fees as consumers repay their loans (.5-1% of outstanding loans).
Historically, $UPST served the personal loan market. From April 2019-March 2020, $118B in personal loans were originated, of which $UPST facilitated $3.5B. Recently, $UPST entered the auto loan market, a much larger opportunity with $625B of auto loans from April 2019-March 2020.
While the top 10 banks may develop AI models on par with $UPST, that is not the target market. There are 5200 banks in the US that hold ~$8 trillion. $UPST aims to partner with small-to-mid-sized banks that lack the resources to invest heavily in an AI lending platform.
Since its founding in 2012, $UPST has given over 620K loans with loan originations increasing over time from 136K for the 9 months ended September 2019 to 177K for the 9 months ended September 2020. Revenue for the same period increased from $101.6M to $146.7M (or up 44%).
$UPST is the rare recent IPO to report three of its four last quarters were profitable on a GAAP basis. Profitably has been achieved in considerable part through AI-based efficiencies. 70% of $UPST’s loans are fully automated, which means there are no to low labor costs.
While $UPST is off to a strong start, competition will likely only become more intense. Like many AI companies, $UPST claims to have a competitive advantage because it’s AI lending model improves with data over time, but the strength of this edge remains to be seen.
Nonetheless, $UPST has plenty of room to grow in its current markets and has other opportunities for expansion such as credit and mortgage markets. Down the road, $UPST may also be an attractive acquisition target for a large bank looking to step up its tech.
$UPST has performed well throughout the pandemic and its outlook will only improve as the US continues to reopen. While the valuation is high, this could also be an opportunity to get in on the next wave of consumer lending.
