Upstart

Prospectus Summary

PROSPECTUS SUMMARY This summary highlights selected information that is presented in greater detail elsewhere in this prospectus. This summary does not contain all of the information you should consider before investing in our common stock. You should read this entire prospectus carefully, including the sections titled “Risk Factors” and “Management’s Discussion and Analysis of Financial Condition and Results of Operations” and our consolidated financial statements and the related notes included elsewhere in this prospectus, before making an investment decision. Unless the context otherwise requires, the terms “Upstart,” “the company,” “we,” “us” and “our” in this prospectus refer to Upstart Holdings, Inc. and its consolidated subsidiaries. Our fiscal year end is December 31, and our fiscal quarters end on March 31, June 30, September 30 and December 31.

Upstart Holdings, Inc.

Overview

Our mission is to enable effortless credit based on true risk.

We are a leading, cloud-based artificial intelligence lending platform. Artificial intelligence, or AI, lending enables a superior loan product with improved economics that can be shared between consumers and lenders. Our platform aggregates consumer demand for high-quality loans and connects it to our network of Upstart AI-enabled bank partners. Consumers on our platform benefit from higher approval rates, lower interest rates, and a highly automated, efficient, all-digital experience. Our bank partners benefit from access to new customers, lower fraud and loss rates, and increased automation throughout the lending process. Since inception, our bank partners have originated over 620,000 personal loans that have generated more than 9 million repayment events. In the nine months ended September 30, 2020, approximately 70% of Upstart-powered loans were entirely automated.

Credit is a cornerstone of the U.S. economy, and access to affordable credit is central to unlocking upward mobility and opportunity. The FICO score was invented in 1989 and remains the standard for determining who is approved for credit and at what interest rate.¹² While FICO is rarely the only input in a lending decision, most banks use simple, rules-based systems that consider only a limited number of variables. Unfortunately, because legacy credit systems fail to properly identify and quantify risk, millions of creditworthy individuals are left out of the system, and millions more pay too much to borrow money.¹³ The first generation of online lenders focused on bringing credit online. Analogous to earlier internet pioneers, these companies made shopping for and accessing credit simpler and easier for consumers and businesses. It was no longer necessary to stand in line at a bank branch, to sit across the desk from a loan officer and to wait weeks or months for a decision. These lenders enabled the emergence of personal loan products that were previously unprofitable for banks to offer. While they brought the credit process online, they inherited the decision frameworks that banks had used for decades and did not address the more rewarding and challenging opportunity of reinventing the credit decision.

We leverage the power of AI to more accurately quantify the true risk of a loan. Our AI models have been continuously upgraded, trained and refined for more than eight years. We have discrete AI

models that target fee optimization, income fraud, acquisition targeting, loan stacking, prepayment prediction, identity fraud and time-delimited default prediction. Our models incorporate more than 1,600 variables and benefit from a rapidly growing training dataset that currently contains more than 9 million repayment events.¹⁴ The network effects generated by our constantly improving AI models provide a significant competitive advantage—more training data leads to higher approval rates and lower interest rates at the same loss rate.

We have been able to demonstrate through several studies that AI lending works. First, in 2019 the Consumer Financial Protection Bureau, or CFPB, reported that a study by Upstart of its data using a methodology specified by the CFPB showed that our AI model approves 27% more borrowers than a high-quality traditional model, with a 16% lower average APR for approved loans.¹⁵ Second, when compared to credit models from several large banks, our AI models approve approximately 2.7 times as many borrowers at the same loss rate.¹⁶ Third, for pools of securitized loans, our realized loss rates were only approximately half of those predicted by Kroll, a prominent credit rating agency; over that same time period, realized losses for the same pool of loans were on average only 5% different than our internal forecasts.¹⁷ And finally, we regularly monitor the accuracy of our AI models in comparison with simple credit score-based models and have observed higher model accuracy across a variety of statistical measures relating to each model’s predictive accuracy.¹⁸ Our AI models are provided to bank partners within a consumer-facing cloud application that streamlines the end-to-end process of originating and servicing a loan. We have built a configurable, multi-tenant cloud application designed to integrate seamlessly into a bank’s existing technology systems. Our highly configurable platform allows each bank to define its own credit policy and determine the significant parameters of its lending program. Our AI models use and analyze data from all of our bank partners. As a result, these models are trained by every Upstart-powered loan, and each bank partner benefits from participating in a shared AI lending platform.

Consumers can discover Upstart-powered loans in one of two ways: either via Upstart.com or through a white-labeled product on our bank partners’ own websites.

Loans issued through our platform can be retained by our originating bank partners, distributed to our broad base of approximately 100 institutional investors and buyers that invest in Upstart-powered loans or funded by Upstart’s balance sheet. In the third quarter of 2020, 22% of the loans funded through our platform were retained by the originating bank and 76% of loans were purchased by

institutional investors through our loan funding programs. Our institutional investors and buyers that participate in our loan funding programs, which include Goldman Sachs, PIMCO and funds managed by Morgan Stanley Investment Management, invest in Upstart-powered loans through whole loan purchases, purchases of pass-through certificates and investments in asset-backed securitizations. We enter into nonexclusive agreements with our whole loan purchasers and each of the grantor trust entities in our asset-backed securitizations, or ABS, under which our ABS investors benefit from our loan servicing capabilities.¹⁹ The remaining 2% of loans funded through our platform in the third quarter of 2020 were funded through our balance sheet.

Our revenue is primarily comprised of fees paid by banks. We charge banks referral fees for each loan referred through Upstart.com and originated by a bank partner, platform fees for each loan originated (regardless of its source) and loan servicing fees as consumers repay their loans. Our agreements with our bank partners are nonexclusive, generally have 12-month terms that automatically renew, subject to certain early termination provisions and minimum fee amounts, and do not include any minimum origination obligation or origination limits. As a usage-based platform, we target positive unit economics on each transaction, resulting in a cash efficient business model that features both high growth rates and profitability. As of September 30, 2020, we had 10 bank partners. In the nine months ended September 30, 2020, Cross River Bank originated 72% of the loans facilitated on our platform and fees received from Cross River Bank accounted for 65% of our total revenue.²⁰ Our current agreement with Cross River Bank began on January 1, 2019 and has an initial four year term, with a renewal term for an additional two years following the initial four year term.

We have achieved rapid growth while improving our margin profile in recent years. The number of loans facilitated on our platform increased by 88% from 114,125 in 2018 to 215,122 in 2019, and 30% from 136,468 in the nine months ended September 30, 2019 to 176,983 over the same period in 2020. Our revenue for the nine months ended September 30, 2019 and 2020 was $101.6 million and $146.7 million, respectively, representing an increase of 44%. For the nine months ended September 30, 2019 and 2020, our net income (loss) was $(6.5) million and $5.0 million, respectively.²¹

Industry Overview

Affordable Credit is Critical to Unlocking Upward Mobility and Opportunity

With $3.6 trillion of consumer credit originated between April 2019 and March 2020,²² credit is a cornerstone of the U.S. economy. Access to affordable credit is central to unlocking upward mobility and opportunity. Reducing the price of borrowing for consumers has the potential to dramatically improve the quality of life for millions of people. Studies have demonstrated a strong statistical link among access to affordable credit, personal well-being and income growth.²³ The average American has approximately $29,800 in personal debt.²⁴ While access to affordable credit has allowed Americans to purchase and improve their homes, buy cars, pay for college tuition and cover emergency expenses, high interest

rates can negatively impact a consumer’s financial health. The U.S. Federal Reserve reports that on average, 10% of household disposable personal income is spent on debt repayment.²⁵ In addition, 16% of Americans spend 50% to 100% of their monthly income repaying debt.²⁶

Affordable Credit Is Inaccessible for Millions because Existing Systems Fail to Accurately Quantify Risk

The FICO score was invented in 1989 and has not fundamentally changed since that time.²⁷ The FICO score is used by over 90% of lenders to determine who is approved for credit and at what interest rate.²⁸ While FICO is rarely used in isolation, many credit models are simple, rules-based systems. A leading expert found that bank credit models commonly incorporate eight to 15 variables, with the more sophisticated models using as many as 30.²⁹ Unsurprisingly, the world is more complicated than can be represented by these models, so they are limited in their ability to reliably estimate the probability of default.

Many borrowers suffer from the effects of inaccurate credit models. Many are approved for a loan that they ultimately will be unable to repay, negatively impacting both the consumer and the lender. Many others may be declined for a loan that they could have successfully repaid if given the opportunity—again doing harm to both consumer and lender. According to an Upstart retrospective study completed in December 2019, four out of five Americans who have taken out a loan have never defaulted, yet less than half of Americans have access to prime credit.³⁰ Even consumers with high credit scores tend to pay too much for loans because the rates they pay effectively subsidize the losses from borrowers who default.

Banks Will Continue to be at the Forefront of Consumer Lending

Banks have been at the forefront of consumer lending in the U.S. for more than a century. They benefit from long-term structural advantages, including a low cost of funding, a unique regulatory framework, and high levels of consumer trust. Through large and reliable deposit bases, banks are able to maintain a very low cost of funds—approximately 1% on average.³¹ These cost savings are passed through to borrowers in the form of lower interest rates, a significant competitive advantage over non-depository lending institutions. Banks also benefit from a regulatory framework that allows them to create nation-wide lending programs that are largely uniform. Given these advantages, we believe that a partnership-based bank enablement approach will be more successful than a disruption strategy.

Banks Must Undergo a Digital Transformation to Remain Competitive

The largest four U.S. banks spend an estimated $38 billion on technology and innovation annually.³² These four banks may attempt to build AI lending models over time, once general market acceptance has been achieved. However, outside the largest four banks, there are approximately 5,200 FDIC insured institutions³³ that are at risk of falling behind. Despite holding over $8 trillion in deposits,³⁴ we believe these banks, particularly small to medium-sized banks, have outdated technology and lack the technical resources of larger banks to fund the digitization process.

At the same time, consumers are increasingly seeking digital, personalized and automated experiences.³⁵ A 2017 Bain survey found that approximately 50% of the U.S. population would be comfortable buying financial products from technology companies.³⁶ We believe that as consumers, both young and old, move their financial lives online, small and medium-sized banks will be increasingly ill-equipped to serve them.

We believe that these trends have been accelerated by the COVID-19 pandemic, as the lack of access to physical bank branches has increased the banking industry’s focus on digital capabilities. The performance of our platform through this crisis has also given existing and prospective bank partners an important new data point to underpin their growing confidence in our solution.

Increasing Recognition from Regulators

Many regulators including the Federal Deposit Insurance Corporation, or FDIC, the Office of the Comptroller of the Currency, or OCC, the Federal Reserve and the CFPB increasingly recognize the opportunity to modernize techniques used in lending.³⁷ In December 2019, these agencies issued an inter-agency report in support of the use of alternative data in lending decisions.³⁸ Additionally, in November 2019, the CFPB director noted that despite external uncertainty regarding how AI will fit into

regulatory frameworks, the CFPB is focused on ensuring a path to regulatory clarity because it recognizes the value AI lending products can offer consumers.³⁹ In fact, in 2017, in response to a request by Upstart, the CFPB issued Upstart the first no-action letter, which provides that the CFPB has no present intention to recommend enforcement action with regard to the application of the Equal Credit Opportunity Act against Upstart for its use of alternative variables and AI and machine learning in credit decision-making.⁴⁰ Such no-action letter expires on December 1, 2020. We are currently discussing with the CFPB whether we can continue to operate under its no-action letter policies, which were revised in 2018 and replaced the prior policy under which our no-action letter was issued. However, there is no guarantee that these discussions will be successful.

The AI Opportunity

AI has the potential to add $13 trillion to the current global economic output by 2030, a 16% increase over today’s output.⁴¹ According to the McKinsey Global Institute, AI will be slowly adopted in its early stages, followed by steep acceleration as the technology matures and companies learn how to best deploy it.⁴² We believe the lending industry will follow this path.

Lending is a compelling application for AI. First, it involves sophisticated decisioning for events that occur millions of times each day. Second, there is an almost unlimited supply of data that has the potential to be predictive and improve the accuracy of credit decisions. Third, given the costs and risks associated with lending, the economic wins from AI are dramatic for both banks and consumers. This means that the significant investment required to overcome the technical and regulatory hurdles is well worth the effort.

With our eight-year head start, our AI lending platform is well-positioned to power a significant portion of the U.S. credit market. To date, we have focused on the unsecured personal loan market, one of the fastest-growing segment of consumer credit.⁴³ From April 2019 to March 2020, there were $118 billion⁴⁴ in U.S. unsecured personal loan originations, representing 8% growth over the prior year. In the same period, we facilitated the origination of $3.5 billion in unsecured personal loans, or less than 5% of the total market.⁴⁵ We not only have a large opportunity to capture market share in unsecured personal loans, but by applying our AI models and technology to adjacent opportunities, we believe we are well-positioned to address the U.S. auto loan, credit card and mortgage markets. From April 2019 to March 2020, there were $625 billion in U.S. auto loan originations, $363 billion in U.S. credit card originations and $2.5 trillion in U.S. mortgage originations.⁴⁶ In June 2020, we began offering auto loans on our platform, and in September 2020, the first auto loan was originated through the Upstart platform. Over time, we believe we are also capable of capturing market share in student loans, point-of-sales loans and Home Equity Lines of Credit, or HELOCs.

Our AI Lending Platform

Our AI models are central to our value proposition and unique position in the industry. Our models incorporate more than 1,600 variables, which are analogous to the columns in a spreadsheet. They have been trained by more than 9 million repayment events, analogous to rows of data in a spreadsheet. Interpreting these almost 15 billion cells of data are increasingly sophisticated machine learning algorithms that enable a more predictive model.

These elements of our model are co-dependent; the use of hundreds or thousands of variables is impractical without sophisticated machine learning algorithms to tease out the interactions between them. And sophisticated machine learning depends on large volumes of training data. Over time, we have been able to deploy and blend more sophisticated modeling techniques, leading to a more accurate system. This co-dependency presents a challenge to others who may aim to short-circuit the development of a competitive model. While incumbent lenders may have vast quantities of historical repayment data, their training data lacks the hundreds of columns, or variables, that power our model. For more details regarding the variables, training data, and algorithms in our models, please see “Business—Evolution of Upstart’s AI Model.”

Despite their sophistication, our AI models are delivered to banks in the form of a simple cloud application that shields borrowers from the underlying complexity. Additionally, our platform allows banks to tailor lending applications based on their policies and business needs. Our bank partners can configure many aspects of their lending programs, including factors such as loan duration, loan amount, minimum credit score, maximum debt-to-income ratio and return target by risk grade. Within the construct of each bank’s self-defined lending program, our platform enables the origination of conforming and compliant loans at a low per-loan cost.

Our platform benefits from powerful flywheel effects that drive continuous improvements as our business scales. Our platform benefits first from increasingly sophisticated models, variable expansion and rapid growth of training data. Upgrades to our platform allow us to offer higher approval rates and lower interest rates to consumers, which increases the number of borrowers on our platform. Upgrades to our platform also lead to better borrower selection, which lowers losses and lowers interest rates to borrowers. The flywheel effect created by self-reinforcing AI increases the economic opportunity that can be shared by borrowers and lenders over time.

Upstart’s AI Flywheel

Our Ecosystem

Our platform connects consumers, banks and institutional investors through a shared AI lending platform. Because AI is a new and disruptive technology, and banking is a traditionally conservative industry, we have brought our technology to market in a way that allows us to grow rapidly and improve on our AI models, while allowing banks to take a prudent and responsible approach to assessing and adopting our platform.

On the consumer side, we aggregate demand on Upstart.com, where consumers are presented with bank-branded offers from our bank partners. In this way, we benefit banks who have adopted our AI lending technology. Bank partners can also offer Upstart-powered loans through a white-labeled interface on their own website or mobile application. Consumers on our platform are generally offered unsecured personal loans ranging from $1,000 to $50,000 in size, at APRs typically ranging from approximately 6.5% to 35.99%, with terms typically ranging from three to five years, with a monthly repayment schedule and no prepayment penalty.

On the funding side, our bank partners can retain loans that align with their business and risk objectives, while the remainder can be sold to our network of institutional investors, which have far broader and more diverse capacity to absorb and distribute risk. This flexible approach allows banks to adopt AI lending at their own pace, while we continue to grow and improve our platform.

Upstart’s Ecosystem

Value Proposition to Consumers

Value Proposition to Bank Partners

Our Competitive Strengths

Constantly Improving AI Models

We have been building and refining our AI models for more than eight years, and they have led directly to our growth and profitability. Our models currently incorporate more than 1,600 variables and are trained by more than 9 million repayment events. Beyond the advantages accrued by our constantly growing volume of training data, our data science team continues to update our modeling

techniques regularly. Model and technology improvements have increased our conversion rate from the initial rate inquiry to funded loan by a factor of ten since inception. We have a pipeline of potential model improvements that we expect will further increase our conversion rates in the future.

Flexible Two-Sided Ecosystem

We benefit from aggregating consumer demand on Upstart.com, referring consumers directly to our network of AI-enabled bank partners. Our consumer presence allows us to increase awareness of and interest in Upstart-powered loans, directly contributing to our own growth, as well as the growth and success of our bank partners’ lending programs.

With an expanding list of bank partners, we can solve the borrowing needs of an increasingly diverse array of consumers. As more banks leverage the Upstart platform, consumers benefit from better offers of credit, while experiencing a consistently high-quality experience.