Upstart's long term competitive advantage?

What do all of you view as Upstart’s long term competitive advantage?

Some thoughts:

Upstart’s success on a company is predicated on it’s AI model’s performance relative to FICO, and to the other currently available credit lending methods. If Upstart is unable to have a superior AI model, they will likely go out of business and lose market share.

  1. Upstart’s competition for best credit model can be grouped in 2 phases:
  • Upstart Model vs. FICO score ( current phase)
  • Upstart Model vs. AI competitors ( entering this one, between the 2 )

Phase 1: How will Upstart get an advantage over FICO ?

  • AI technology improvement - AI training costs and model improvements are falling dramatically
  • AI training costs are falling 50x faster than Moore’s law!! - Ark Invest
    " In 2017, for example, the cost to train an image recognition network like ResNet-50 on a
    public cloud was ~$1,000. In 2019, the cost dropped to ~$10, as shown below. At the current
    rate of improvement, the cost should fall to $1 by the end of this year.[2] The cost of
    inference—running a trained neural network in production—has dropped even more precipitously.
    During the past two years, for example, the cost to classify one billion images has fallen from
    $10,000 to just $0.03, as shown below.[3] "
  • https://ark-invest.com/articles/analyst-research/ai-training…
  • This means that Upstart( and the competition) can iterate much quicker at a far lower cost
    than before. This should accelerate their model performance improvements, while reducing costs
  • AI performance should scale well with more compute power + more data
  • The more data the AI gets the better it tends to do, given it has the compute to match with
    that. - Andrew NG - “AI is the new electricity” https://youtu.be/21EiKfQYZXc
  • Upstart model performance will likely get better the more data they get.

Phase 2: how will Upstart defend it’s model superiority vs. AI competitors over time?

  • Maybe a positive network effect?
  • Each bank partner that Upstart adds loan origination volume to Upstart’s network which makes
    Upstart’s credit models better ( more data ) which increases the value of Upstart’s network
    which makes even more bank partners join.
  • Additionally, there is little incentive for a bank to leave Upstart’s network unless the
    competition makes an exceptionally good AI model. There would be a switching cost in terms of
    degraded performance to going to a competitors solution.
  • More importantly: This network effect is reversed for upstart’s competitions
  • The better Upstart’s models become → the more bank partners leave to join upstart’s
    platform → The less data the competition’s models have → The less improvements they
    are able to make → The more banks go to Upstart ( better loans) → The better
    Upstart’s models become.
  • A death spiral for the competition
  • Buying out the competition
  • Upstart has the capital to buy out early stage competition
  • Regulators will probably not like this.

The next post will go into some risks with these moats.

23 Likes
Disclosure before I get into the risks - Currently Long Upstart. 

These thing's won't necessarily happen, just some items to watch out for. 

Some Risks: 

* Upstart's data isn't enough to escape competition 
     - The Data Upstart has isn't enough unique enough: 
           " As you gather data, the data also tends to become less valuable to add to the corpus. 
             Why? Even if the new arbitrary batch of data has the same cost to collect as the last 
             batch acquired, it yields less value given some of the new data you acquire already 
             overlaps with your existing corpus. And this only gets worse over time: Benefits of new 
             data go down. " 
           - [https://a16z.com/2019/05/09/data-network-effects-moats/](https://a16z.com/2019/05/09/data-network-effects-moats/) 
           - This quotation highlights that data diversity and breadth is essential. Maybe this is
             another reason why Upstart's loan delinquencies are increasing, besides more less-prime 
             borrowers + changing fiscal environment . these events would help them collect more data
             , and help expand there credit-score range even further. 
                    - In this sense, delinquent loans are an incredibly valuable data set. 
                         - Upstart know has a great data set on who NOT to make loans to. 
           - If Upstart doesn't get enough unique data, they might not be able to withstand a 
             competitive threat. 
     - Industry wide Model improvements make the Data advantage less significant: 
         - A recent example: 
         - " .....language model Chinchilla (70B) outperforms much larger Gopher (280B), GPT-3 
           (175B), Jurrasic-1 (178B), MT-NLG (530B)  [https://arxiv.org/abs/2203.15556](https://arxiv.org/abs/2203.15556) Important new 
           LM scaling laws paper from DeepMind. Go smaller, train longer. Many misconfigurations 
           likely continue to lurk. " Andrej Karpathy twitter 
         - Note there is a factor of 2 difference in model size, yet it was still able to outperform 
           the larger models. 
         - It will be important to match Upstart's data to their competitors over time. They are probably safe 
           as long as they are greater than a factor of at least 2 apart in terms of data. 
* An underlying performance threshold is reached  ( 1 ) 
         - Upstart's model performance hits a performance threshold; competition eventually 
           catches up and AI lending becomes a low-margin commodity product.  
(1)- [https://techcrunch.com/2018/03/27/data-is-not-the-new-oil/](https://techcrunch.com/2018/03/27/data-is-not-the-new-oil/) 

Data moats have failed historically.
1. Amazon's Alexa outpaced Siri, and then started losing market share to Google - (90+% to less than 30% ) 
[https://venturebeat.com/2019/02/10/data-is-not-the-new-oil/](https://venturebeat.com/2019/02/10/data-is-not-the-new-oil/) 
[https://www.statista.com/statistics/792604/worldwide-smart-s...](https://www.statista.com/statistics/792604/worldwide-smart-speaker-market-share/)
 " The same story has played out time and time again. Flight data collected by drone maker DJI 
   hasn’t kept it safe from Skydio, which devised better algorithms for avoiding obstacles. Uber’s 
   bumper crop of data about drivers, passengers, and routes hasn’t fended off Lyft. Facebook, even 
   with snapshots spanning nearly a third of humanity, had to buy Instagram to neutralize an 
   existential threat. This isn’t new: Yahoo, which in 1998 had more web-search data than anyone, 
   got crushed by then-upstart Google.([https://venturebeat.com/2019/02/10/data-is-not-the-new-oil/](https://venturebeat.com/2019/02/10/data-is-not-the-new-oil/))
        - Perhaps it isn't a fare comparison given AI has progressed so much since then, but the 
underlying point is still there.  

Some additional elements that might give more context on the defensibility of Upstart's data advantage/disadvantage: 
* Accessibility - How easy is this data to collect? 
* Time - How quickly can the data be amassed 
* Cost - How much money is needed to acquire this data? 
* Uniqueness - is similar data widely available? 
* Dimensionality 
* Breadth - How widely do the values of attributes vary? 
* Perishability-how broadly applicable over time is the data?
* Virtuous Loop 
([https://techcrunch.com/2018/03/27/data-is-not-the-new-oil/](https://techcrunch.com/2018/03/27/data-is-not-the-new-oil/) ) 

I think Upstart will probably have a good advantage in the breadth and dimensionality of their data. 
Their data also has very low acquisition costs. In fact, they are PAYED to get data. Upstart should
 also have the culture to use this data. ([https://dzone.com/articles/data-moats-are-not-just-about-](https://dzone.com/articles/data-moats-are-not-just-about-)
the-data ) 

More good articles: 
[https://dzone.com/articles/data-moats-are-not-just-about-the...](https://dzone.com/articles/data-moats-are-not-just-about-the-data) 
[https://creativeventures.vc/2021/01/14/the-fall-of-data-moat...](https://creativeventures.vc/2021/01/14/the-fall-of-data-moats/)
AI consolidation: [https://twitter.com/karpathy/status/1468370605229547522?s=20...](https://twitter.com/karpathy/status/1468370605229547522?s=20&t=ZXBspBwd40DVGGZdA5lrhw)

Good Interview: [https://ir.upstart.com/events/event-details/jmp-securities-t...](https://ir.upstart.com/events/event-details/jmp-securities-technology-conference)
13 Likes

str1dr - This is a very good, fair and balanced post. Thanks.

I want to add one negative and one positive point.

  • The AI problem Upstart is trying to solve is NOT difficult. “Self driving car” or “Cancer detection” is a hard problem. What Upstart has, a bunch of smart ML scientists can build (“IF” they have access to the data)
  • Upstart has the inbuilt flywheel built in more banks → more data → better models → more banks.
    This - brand, trust, integrations with banks - is a relatively hard problem for someone is a garage or even a large bank to solve.

I would like Upstart to land a “whale”, which they eventually will. That would be a big inflection point.

8 Likes

I agree with dividends20: good and useful posts by Str1der. And I’ll just add a little more about part of Upstarts long term moat that divi touched on: brand and trust. Divi mentioned this as related to their relationship with banks:

+ Upstart has the inbuilt flywheel built in more banks → more data → better models → more banks.
This - brand, trust, integrations with banks - is a relatively hard problem for someone is a garage or even a large bank to solve.

All very true, and an important and potentially durable advantage for Upstart. But this also applies to Upstart’s direct relationships with loan consumers. Upstart has a two pronged attack to generating business: their partnerships with banks, credit unions, auto dealers, etc … where that second party rebrands Upstart’s software interface as their own, but Upstart also pursues a direct relationship with loan consumers as an internet brand of their own at upstart.com where right now you can shop for personal and auto loans.

Assuming their direct customers have a good experience, they will refer other customers and become repeat customers themselves, possibly for future products like home mortgages too. A well-liked and trusted consumer brand can be a good moat for decade after decade, so long as the brand continues to live up to its promises, even if in the future another company offers an equivalent interface and AI for loan application and qualification, if Upstart has a multitude of loyal direct customers, they will be able to retain them and even charge a premium over the competitor.

Ben ← not a shareholder as yet, but hoping to buy in at a good price.

2 Likes
If Upstart lands a whale, they probably won't make any profit from it since the big bank has so much leverage. But the data will be incredible. 

If you are Jamie Dimon: Getting Upstart's model for free + giving them training data is probably less expensive than hiring an internal development effort which may catch up. 
 I would be very happy if Upstart gave their model for 80-90% off to the big banks. 

A case study of Upstart V. Competition: 

LP - Lending point ( [https://www.lendingpoint.com/](https://www.lendingpoint.com/) ) 
FREED - " India's top debt relief platform" -> [https://freed.care/](https://freed.care/) 
This is from KBRA.
( [https://www.kbra.com/documents/report/62312/upstart-securiti...](https://www.kbra.com/documents/report/62312/upstart-securitization-trust-2022-1-abs-kcat)) 

Deal Name	                        UPST 2022-1	LP 2022-A	FREED 2022-1FP
Transaction Date	                 3/30/2022*	1/26/2022	1/26/2022
                                           Collateral Stratification			
Pool Balance (as of cutoff date)	$503,736,059	$264,011,540	$273,988,098
Number of Loans 	                 65,724	        26,819	         13,268
Avg Loan Balance	                 $7,664	        $9,844	          $20,650
Wtd Avg Coupon	                         18.77%	        20.60%	           16.17%
Wtd Avg FICO	                         654	          664	             698
Wtd Avg Original Term (mths)	         56	           50	              52
Wtd Avg Remaining Term (mths)	         52	           49	              49
Wtd Avg Seasoning (mths)	          4	           1	              3
                                                   FICO Distribution			
619 and Lower	                        30.48%	         13.54%	              --
620 to 639	                        13.14%	         13.59%	              --
640 to 659	                        14.92%	         19.17%	              --
660 to 679	                        12.21%	         19.22%	         551 to 600: 0.02%
680 to 699	                        9.01%	         16.12%	         601 to 650: 12.66%
700 to 719	                        5.78%	         10.88%	         651 to 700: 44.26%
720 to 739	                        6.78%	         5.41%	         701 to 750" 30.94%
740 to 759	                        3.80%	      740+: 2.06%	 751 to 800: 10.00%
760 to 779	                        2.02%	          --	            801+: 2.12%
780 to 799	                        1.13%	          --	                --
800 to 819	                        0.58%	          --	                --
820 and Higher	                        0.15%	          --	                --
                                            Geographic Concentration			
State 1	                              CA: 13.12%	CA: 12.04%	CA: 13.89%
State 2	                              TX: 11.29%	TX: 8.32%	TX: 12.53%
State 3	                              NY: 6.47%	        FL: 8.32%	FL: 8.49%
Note Balance (* Not relevant. Included in the document. )			
                                                 % Credit Enhancement (1)			
Initial O/C	                         9.00%	         5.00%	               15.00%
Target O/C (3)	                         9.00%	         6.00%	               21.50%
O/C Floor	                         2.00%	         2.00%	               2.00%
Reserve Account                        	 0.50%	         0.50%	                0.00%
                                                  Gross Excess Spread			
Collateral Interest Rate	         18.77%	        20.60%	                16.17%
Note Coupon (Weighted)(2)	         4.14%	        3.44%	                2.37%
Servicing Fees	                         0.80%	        1.00%	                1.00%
Total Gross Excess Spread	         13.83%	        16.16%	                12.79%
                                             % Total Initial Credit Enhancement			
AAA (sf) Class	                          --	          --	                58.75%
AA- (sf) Class                            --	        44.50%	                37.75%
A- (sf) Class	                        33.50%	        34.60%	                28.25%
BBB- (sf) Class 	                21.97%	        25.20%	                15.00%
BB- (sf) Class	                        9.50%	        9.50%	                  --
B (sf) Class	                         --	        5.50%	                  --
                                    KBRA Base Case Cumulative Net Loss Expectation	 		
KBRA Base Case Loss Range	   14.50% - 16.50%   13.95% - 15.95%	    12.05% - 14.05%

Some takeaways: 
* Loss rate is slightly worse than FREED. 
        - 14.50 base loss rate vs. 12.05% for FREED. 2% Spread. BUT 
        - Upstart has about a 50 pt lower FICO score (* However: Loss rate increasing as FICO decreased was not a part of Upstart's original value proposition) 
        - Upstart pool size is 200 million dollars bigger. 500 million vs. 272 million. It Will be interesting to see how FREED's model's perform as they sale.  
* Upstart should be collecting significantly more repayment data 
        - each origination and performance of that loan is an event that will likely get fed back into upstart's AI models 
        - Upstart originated: 65,724 distinct loans vs. FREED originating 13,628 loans. 
               - 4x the number of loan repayment events from this pool than FREED. 
               - So upstart might be able to iterate it's models much faster. 
               - This Should bode well for future model improvements. 
        - More broad loan events pool 
               - UPST  did 65,724 loans with a weighted average FICO of 654
               - FREED did 13,268 loans with a  weighted average FICO of 698 
               - BUT. Looking through the lens of repayment events 
                     - Comparing Upstart loans below 640 Vs. FREED Below 650 
                     - UPST did  43.62% of there loans below a 640 FICO score 
                     - FREED did 12.68% of there loans below a 650 FICO score
                     - So Upstart did: 28,668 Loans below a 640 FICO Score 
                     -      FREED did:  loans below a 650 FICO score
                           - A factor of 10 difference!!
                     - Multiplying by WTD average loan amount 
                          * Upstart originated  28,668 loans * 7664 $/loan which is  about 219,700,000$ 
                          *   FREED originated 1682.38 loans * 20,650 $/loan which is about 34,741,000$ 
                          * More than a factor of 6 difference  
                          * These  numbers will both likely be lower given the WTD average FICO is a lot higher. This means they gave larger dollar amounts to higher FICO borrowers.
 This Serves to illustrate the difference in loan data they are gathering. 

Do you think the loan amount is significant enough to get a true representation? Is 5000$ loan amount enough? 
You could imagine giving a 10$ loan to 500,000 borrowers with FICO< 500, getting 100% repayments, 
and then having your algorithm blowing up as you are trying to scale. The 10$ loan wasn't a 
significant enough loan. I think I will need to trust management that they are taking this into consideration. 

- 
str1der
4 Likes

Reposting Str1der’s post, formatted for readability.

If Upstart lands a whale, they probably won’t make any profit from it since the big bank has so much leverage. But the data will be incredible.

If you are Jamie Dimon: Getting Upstart’s model for free + giving them training data is probably less expensive than hiring an internal development effort which may catch up.
I would be very happy if Upstart gave their model for 80-90% off to the big banks.


A case study of Upstart V. Competition: 

LP - Lending point ( [https://www.lendingpoint.com/](https://www.lendingpoint.com/) ) 
FREED - " India's top debt relief platform" -> [https://freed.care/](https://freed.care/) 
This is from KBRA.
( [https://www.kbra.com/documents/report/62312/upstart-securiti...](https://www.kbra.com/documents/report/62312/upstart-securiti...)) 

Deal Name	                        UPST 2022-1	LP 2022-A	FREED 2022-1FP
Transaction Date	                 3/30/2022*	1/26/2022	1/26/2022
                                           Collateral Stratification			
Pool Balance (as of cutoff date)	$503,736,059	$264,011,540	$273,988,098
Number of Loans 	                 65,724	        26,819	         13,268
Avg Loan Balance	                 $7,664	        $9,844	          $20,650
Wtd Avg Coupon	                         18.77%	        20.60%	           16.17%
Wtd Avg FICO	                         654	          664	             698
Wtd Avg Original Term (mths)	         56	           50	              52
Wtd Avg Remaining Term (mths)	         52	           49	              49
Wtd Avg Seasoning (mths)	          4	           1	              3
                                                   FICO Distribution			
619 and Lower	                        30.48%	         13.54%	              --
620 to 639	                        13.14%	         13.59%	              --
640 to 659	                        14.92%	         19.17%	              --
660 to 679	                        12.21%	         19.22%	         551 to 600: 0.02%
680 to 699	                        9.01%	         16.12%	         601 to 650: 12.66%
700 to 719	                        5.78%	         10.88%	         651 to 700: 44.26%
720 to 739	                        6.78%	         5.41%	         701 to 750" 30.94%
740 to 759	                        3.80%	      740+: 2.06%	 751 to 800: 10.00%
760 to 779	                        2.02%	          --	            801+: 2.12%
780 to 799	                        1.13%	          --	                --
800 to 819	                        0.58%	          --	                --
820 and Higher	                        0.15%	          --	                --
                                            Geographic Concentration			
State 1	                              CA: 13.12%	CA: 12.04%	CA: 13.89%
State 2	                              TX: 11.29%	TX: 8.32%	TX: 12.53%
State 3	                              NY: 6.47%	        FL: 8.32%	FL: 8.49%
Note Balance (* Not relevant. Included in the document. )			
                                                 % Credit Enhancement (1)			
Initial O/C	                         9.00%	         5.00%	               15.00%
Target O/C (3)	                         9.00%	         6.00%	               21.50%
O/C Floor	                         2.00%	         2.00%	               2.00%
Reserve Account                        	 0.50%	         0.50%	                0.00%
                                                  Gross Excess Spread			
Collateral Interest Rate	         18.77%	        20.60%	                16.17%
Note Coupon (Weighted)(2)	         4.14%	        3.44%	                2.37%
Servicing Fees	                         0.80%	        1.00%	                1.00%
Total Gross Excess Spread	         13.83%	        16.16%	                12.79%
                                             % Total Initial Credit Enhancement			
AAA (sf) Class	                          --	          --	                58.75%
AA- (sf) Class                            --	        44.50%	                37.75%
A- (sf) Class	                        33.50%	        34.60%	                28.25%
BBB- (sf) Class 	                21.97%	        25.20%	                15.00%
BB- (sf) Class	                        9.50%	        9.50%	                  --
B (sf) Class	                         --	        5.50%	                  --
                                    KBRA Base Case Cumulative Net Loss Expectation	 		
KBRA Base Case Loss Range	   14.50% - 16.50%   13.95% - 15.95%	    12.05% - 14.05%

Some takeaways:

  • Loss rate is slightly worse than FREED.
  • 14.50 base loss rate vs. 12.05% for FREED. 2% Spread. BUT
  • Upstart has about a 50 pt lower FICO score (* However: Loss rate increasing as FICO decreased was not a part of Upstart’s original value proposition)
  • Upstart pool size is 200 million dollars bigger. 500 million vs. 272 million. It Will be interesting to see how FREED’s model’s perform as they sale.
  • Upstart should be collecting significantly more repayment data
  • each origination and performance of that loan is an event that will likely get fed back into upstart’s AI models
  • Upstart originated: 65,724 distinct loans vs. FREED originating 13,628 loans.
  • 4x the number of loan repayment events from this pool than FREED.
  • So upstart might be able to iterate it’s models much faster.
  • This Should bode well for future model improvements.
  • More broad loan events pool
  • UPST did 65,724 loans with a weighted average FICO of 654
  • FREED did 13,268 loans with a weighted average FICO of 698
  • BUT. Looking through the lens of repayment events
  • Comparing Upstart loans below 640 Vs. FREED Below 650
  • UPST did 43.62% of there loans below a 640 FICO score
  • FREED did 12.68% of there loans below a 650 FICO score
  • So Upstart did: 28,668 Loans below a 640 FICO Score
  • FREED did: loans below a 650 FICO score
  • A factor of 10 difference!!
  • Multiplying by WTD average loan amount
  • Upstart originated 28,668 loans * 7664 $/loan which is about 219,700,000$
  • FREED originated 1682.38 loans * 20,650 $/loan which is about 34,741,000$
  • More than a factor of 6 difference
  • These numbers will both likely be lower given the WTD average FICO is a lot higher. This means they gave larger dollar amounts to higher FICO borrowers.
    This Serves to illustrate the difference in loan data they are gathering.
5 Likes

This was a very interesting discussion and thanks for starting the conversation, Str1der. Always a good exercise to think of the bear cases for our favorite stocks so that we stay vigilant to the downside possibilities and don’t get too drunk on all the positive Koolaid.

First, I wanted to share this podcast where they interviewed Dave Girouard, CEO of Upstart. Very enlighening chat about the company’s origins, culture, hiring approach, mission etc.
https://podcasts.apple.com/us/podcast/the-business-brew/id15…

Secondly, I agree with the possible risks mentioned above. Each of them could happen and could negatively impact UPST’s business. However I don’t expect UPST to keep swimming in a straight line as they have publicly shared…meaning, from personal loans to auto loans to mortgage loans etc…essentially staying in the loan origination lane.

Once you have the data (millions of attributes), talent (expertise) and a set of functional AI models (experience), I would expect them to look for entry points into adjacent business use cases. This is the Google way, where Dave G earned his stripes.

Off the top of my head, here are some businesses that need to process tons of demographic data in a short period of time to make a transactional decision (not a comprehensive list):

  1. Insurance - auto, home, business, health etc.
  2. Credit cards
  3. Background checks
  4. Travel and immigration checks
    I am sure there are a few more to add here…and each of these are massive TAMs and ripe for disruption.

UPST is, imo, the most interesting & innovative fintech stock to own today. That said, the market is struggling to value the stock and macro conditions (interest rates, consumer sentiment, war, inflation etc.) are taking their toll on its share price.

Beachman (beachman.substack.com)

47 Likes

What’s great about starting with personal loans is that they are learning the most fundamental case of lending: accurately determining whether a person can repay a loan. (from an Interview with their management.

Once you introduce an asset, it might even become EASIER to price the loan. Especially if the asset should appreciate over time.

Judging by their annualized market share, they are becoming the dominant player in personal loans.

3 Likes

Beachman, I would like to affirm your observation of the massive opportunity UPST has to expand the use cases for their numerous capabilities:

-developing useful and effective AI/ML and thereby creating a moat
-navigating significant regulatory constraints and thereby creating more moat
-managing digital fraud threats thereby creating more moat
-collecting huge amount of data while making money
-ability to identify and exploit market inefficiencies
-proven ability to execute
-developing lower friction consumer experiences
-effectively and quickly integrating a recent acquisition
-constant development of multiple new products
-effectively partnering with a wide variety of stakeholders
-making those partnerships rewarding for all stakeholders
-maintaining humility and learning rapidly in a very dynamic space
-hiring great and rapidly growing numbers of talented employees
-maintaining access to huge and growing amounts of data
-operating in a new and risky area while effectively limiting credit risks to the company

You mentioned credit cards as a possible future line of business.

The most recent Leaders in Lending Podcast posted yesterday, and apparently recorded in February, included an interesting player in the credit card business.

https://www.youtube.com/watch?v=1PTOB3j7Auw

Bank executive,Jerry O’Flanagan, EVP at First National Bank of Omaha, a bank with a significant credit card business was interviewed. Topics included

  • Building and maintaining a team for a credit program
  • The impact of buy now pay later
  • Gaining inspiration for innovation from fintechs
  • The strategy behind standing up a national digital bank

What stands out to me about these podcasts is that they show that UPST is having conversations with a variety of significant stakeholders all the time AND LEARNING FROM THEM. They will continue to innovate and adapt. Sure the next products in the pipeline (personal, auto, mortgage, micro, smb) are important (and perhaps already priced in to the stock). However, the stock price appreciation will come from the expanding TAM which is likely to be competently and profitably served by a proven leadership team! As the base of products matures and becomes more predictable, the revenue stream will become less lumpy and we may see the multiple expand once again.

UPST leadership has one limitation that the CEO has acknowledged. He is not that good at raising money from investors! During UPST’s early days this forced the company to become lean and capable in order to survive. Imagine what would happen to the stock price if Snowflake’s Frank Slootman, Cloudfare’s Mathew Prince, Datadog’s Olivier Pomel, or Crowdstrike’s George Kurtz were hyping the future possibilities of the company every quarter! This lack of stock hype has created a buying opportunity for all of us as UPST continues to grow at a rapid pace. In fact, I am much more comfortable owning UPST now that some of the hype has passed.

While not SaaS, I believe that UPST possesses crucial attributes of many great companies; e.g. innovation, talent, founder led, demonstrated profitability, proven ability to execute, and a head start in a greenfield opportunity. And, its stock is on sale now.

While (as Saul regularly states) I have no idea what will happen next, I am thrilled to be able to own a small piece of UPST as the future unfolds! I expect UPST stcok to be worth far more in 3 years. After writing this, I am considering adding to my position!

Good luck to you all!

wishyouwell
Long UPST 12%

23 Likes

“(and perhaps already priced in to the stock)”

I don’t think significant penetration is priced into Upstart stock.

A back of napkin valuation method: Very approximate
Enterprise Value = Ebit * 20
Substitute Net Income for EBIT. ( Probably under ebit by 10-15%. This will underestimate enterprise value )
Enterprise Value = Total Market Volume*Market Share *0.01439726696$/$ of loan originated * 20 EV/EBIT multiple
The 0.01439726696 $/ $ of loan originated was Upstart’s Q4 net income/Q4 $'s originated, the extra digits are significant at larger volumes

Upstart has a total addressable market of 6 trillion in loan originations that will only expand over time.

Upstart’s Enterprise Value vs Market Penetration

Market Penetration (%) EV ($) Upstart Loan Originations ($ )
0.5% 8.63 Billion 30 Billion
1% 17.2 Billion 60 Billion
5% 86.3 Billion 300 Billion
10% 172.7 Billion 600 Billion
20% 345.5 Billion 1.2 Trillion
30% 518.3 Billion 1.8 Trillion
60% 1.03 Trillion 3.6 Trillion
… … …
90% 1.55 Trillion 5.4 Trillion

Upstart has a current Enterprise value of: 8.873 Billion
From the chart it looks like about 1% Penetration is priced into Upstart’s stock longer term.

Upstart has 17.08% Annualized Market Share in Personal Loans. (Q4 *4) / (Personal Loan TAM )

this chart also stresses the importance of Upstart building a long term competitive advantage.

Which of these scenarios is the most reasonable?
1% is very achievable, this would be doubling 2 times from where they are today. Anything above that is a bonus. I don’t think 90% will happen, but it could be a good extreme-bull case.

5 Likes
Reformatting for readability 

Market Penetration (%)    EV ($)           Upstart Loan Originations ($ )
0.5%                   8.63 Billion                30 Billion
1%                     17.2 Billion                60 Billion
5%                     86.3 Billion               300 Billion
10%                    172.7 Billion              600 Billion
20%                    345.5 Billion              1.2 Trillion
30%                    518.3 Billion              1.8 Trillion
60%                    1.03 Trillion              3.6 Trillion
...                    ....                         ....
90%                    1.55 Trillion              5.4 Trillion
2 Likes