Upstart Holdings, Inc. (UPST)

ITEM 2. MANAGEMENT’S DISCUSSION AND ANALYSIS OF FINANCIAL CONDITION AND RESULTS OF OPERATIONS

The following discussion and analysis of our financial condition and results of operations should be read in conjunction with the condensed consolidated financial statements and related notes thereto included elsewhere in this Quarterly Report on Form 10-Q. This discussion contains forward-looking statements that involve risks and uncertainties. Factors that could cause or contribute to such differences include those identified below and those discussed in the section titled “Risk Factors” and other parts of this Quarterly Report on Form 10-Q. Our historical results are not necessarily indicative of the results that may be expected for any period in the future.

Overview

Upstart is the leading AI lending marketplace. We aim to radically reduce the cost and complexity of borrowing for all Americans by using our proprietary AI models to remake the entire lending process. Upstart’s marketplace supports both unsecured and secured loans. Unsecured loans include personal loans, small dollar loans, and our new unsecured revolving credit product, Cash Line. Secured loans include auto loans – including retail, refinance, and auto secured personal loans – and HELOCs. Long-term, our vision is to become the always-on, everything-store for credit, where we can automatically approve borrowers at the right prices – instantly and effortlessly.

Our platform applies AI to more accurately quantify the true risk of a loan, a capability we refer to as “risk separation.” This differentiated approach to underwriting has generally led to higher approvals and lower interest rates relative to traditional lending practices, with more predictable returns to our capital partners including banks and credit unions (collectively our “lending partners”) and institutional investors. With this as the foundation, we’ve added layers of automation, macroeconomic calibration, and personalization that can support increasing scale and greater business resilience over time.

Beyond core underwriting, we apply our proprietary AI models to other areas of our business, such as income and identity verification, fraud detection, and identifying loan stacking behavior, among others. The result is an exceptional digital-first experience with significant levels of automation. For example, during the six months ended June 30, 2026, 91% of loans on our platform were fully automated, with no human intervention by Upstart. Consumer acquisition is another area where we apply our AI, making these activities increasingly efficient. Consumers primarily access Upstart-powered loans through Upstart.com and, for automotive retail in particular, through auto dealerships that use Upstart’s Auto Finance software.

Our dynamic marketplace allows us to serve borrowers across the credit spectrum. Loans issued through our marketplace are purchased by our network of institutional investors, retained or purchased by our lending partners, or in certain instances, held on our balance sheet. Out of the total principal of loans transacted on our marketplace during the six months ended June 30, 2026, 61% were purchased by institutional investors, 31% were retained or purchased by our lending partners, and 8% were held on our balance sheet. Investors may also invest in securities collateralized by Upstart-powered loans through our pass-through and securitization programs.

Institutional investors play an important role in our lending marketplace by providing capital for higher risk loans that may not be economically feasible for traditional banks and credit unions to hold. Today, more than 50% of the loan funding on our platform is through committed capital and other co-investment arrangements with institutional investors and lending partners, which provide valuable stability and resilience on the funding side of our platform.

We retain certain loans on our balance sheet for R&D purposes, including to test and evaluate our AI models for newer products and to aid in price discovery. As of June 30, 2026, 51% of loans held on our balance sheet were for R&D purposes, primarily related to our seasoned auto refinance, auto retail loans and HELOCs, as well as newer products including small dollar loans and auto secured personal loans. As products develop and

65

Table of Contents

Upstart Holdings, Inc.

Management’s Discussion and Analysis of Financial Condition and Results of Operations

(Tabular Amounts in Thousands, Except Share and Per Share Data and Ratios, or as Noted)

achieve market fit, we no longer consider them as R&D. For example, during the quarter ended June 30, 2026, newly originated auto refinance, auto retail loans and HELOCs were no longer classified as R&D.

We may also retain loans on our balance sheet for purposes unrelated to product development, including to bridge the timing between origination and sales of loans.

Credit Performance

We consider credit performance of Upstart-powered loans to be one of the most important measures of the effectiveness of our AI models. However, credit performance is impacted by multiple factors, including factors that our models do not predict, such as macroeconomic conditions.

We evaluate the credit performance of our loans by comparing the target returns expected at the time of origination to the returns received by our lending partners, institutional investors, or us. The target return, a critical component of our loan pricing, is calculated using estimated cash flows, which are developed based on a number of factors, including credit losses and prepayment rates. While target returns across our lending partners and institutional investors vary depending on their programs’ objectives and risk tolerance, overall performance is calculated based on the variance between the initially expected returns and the actual return on capital invested in Upstart-powered loans.

An equal investment in all vintages of Upstart-powered personal loans originated in the second quarter of 2023 through the first quarter of 2026 is currently expected to deliver annual returns in line with a blended target of approximately 10.8% after servicing fees.

At a more granular level, the quarterly vintages originated in the second quarter of 2023 through the first quarter of 2024 and in the fourth quarter of 2024 are currently forecasted to underperform relative to their target returns. Quarterly vintages originated in the second and third quarter of 2024 and in the first quarter of 2025 or later are currently forecasted to deliver returns in line with target yields. This reversion in performance to target yields reflects a combination of factors including increased conservatism in underwriting, the relative stabilization of macroeconomic conditions, and improvements in our more recent models.

Impact of Macroeconomic Environment

In addition to impacting credit performance, the macroeconomic environment has a direct and indirect impact on our business financial condition, and results of operation. In an economic downturn, we believe consumer lending will generally contract. Lending partners and institutional investors will generally require higher rates of return, which in turn increases the interest rates offered to borrowers, leading to lower borrower demand. Macroeconomic factors can also cause fluctuations of available capital in our lending marketplace due to shifts in the risk preferences of our lending partners and institutional investors. We expect these dynamics would generally invert in an economic upswing.

For example, loan funding provided by institutional investors started to become constrained in 2022, largely due to concerns about the macroeconomic environment. Rising interest rates also led to more expensive loan offers across borrower categories, which decreased borrower demand. In order to create greater stability for our business, we began securing committed capital and co-investment arrangements with institutional investors and other third parties that provide loan funding over longer durations. While we believe that the macroeconomic environment started to improve in 2024, disruption in financial markets could once again lower borrower demand or impair our lending partners and result in constrained funding, which would adversely impact our business, financial condition and operating results.

To respond to macroeconomic changes and provide relevant and up-to-date information to our lending partners, we introduced a new metric, the UMI, in 2023. As of June 30, 2026, UMI remained elevated, measured at

66

Table of Contents

Upstart Holdings, Inc.

Management’s Discussion and Analysis of Financial Condition and Results of Operations

(Tabular Amounts in Thousands, Except Share and Per Share Data and Ratios, or as Noted)

approximately 1.50, meaning that current macroeconomic conditions contributed an incremental risk of approximately 50% to the repayment performance of an Upstart-powered unsecured personal loan, compared to the baseline measurement of 1.0.

Key Operating and Non-GAAP Financial Metrics

We focus on several key operating and Non-GAAP financial metrics to measure the performance of our business and help determine strategic direction. The following presents our key operating and financial metrics:

_______

(1)Originations, Number of Loans, is shown in ones for the periods presented.

(2)Beginning in the fourth quarter of 2025, we revised the definition and underlying calculation methodology of Conversion Rate. Prior period figures have been recast to conform to the new definition and methodology. For additional information regarding this change, see “Key Operating and Non-GAAP Financial Metrics” in our Annual Report on Form 10-K for the year ended December 31, 2025.

(3)Represents a non-GAAP financial measure. See the section titled “Management’s Discussion and Analysis of Financial Condition and Results of Operations—Reconciliation of Non-GAAP Financial Measures” for further information.

Originations

Beginning in the second quarter of 2026, we refer to the metrics “Transaction Volume, Dollars” and “Transaction Volume, Number of Loans” as “Originations, Dollars” and “Originations, Number of Loans,” respectively, to reflect management’s internal terminology. We define Originations, Dollars as the aggregate of: (i) the total principal of loan originations for personal loans, small dollar loans, and auto loans, (ii) committed amounts for HELOCs, and (iii) drawn amounts for unsecured revolving credit lines (Cash Line), in each case facilitated on our marketplace during the periods presented. We define Originations, Number of Loans as the total number of such originations, commitments, and draws, as applicable, facilitated on our marketplace during the periods presented. We believe these metrics are good proxies for our overall scale and reach as a marketplace.

For Cash Line, we count each draw, rather than the establishment of the credit line, because draws represent the customer’s use of the product and more closely reflect the marketplace activ

[Excerpt truncated for page length; source filing is linked above.]

Management’s Discussion and Analysis of Financial Condition and Results of Operations

(Tabular Amounts in Thousands, Except Share and Per Share Data and Ratios, or as Noted)

Credit Performance

We consider credit performance of Upstart-powered loans to be one of the most important measures of the effectiveness of our AI models. However, credit performance is impacted by multiple factors, including factors that our models do not predict, such as macroeconomic conditions.

We evaluate the credit performance of core personal loans by comparing the target returns expected at the time of origination to the returns received by our lending partners, institutional investors, or us. The target return, a critical component of our loan pricing, is calculated using estimated cash flows, which are developed based on a number of factors, including credit losses and prepayment rates. While target returns across our lending partners and institutional investors vary depending on their programs’ objectives and risk tolerance, overall performance is calculated based on the variance between the initially expected returns and the actual return on capital invested in Upstart-powered loans.

An equal investment in all vintages of Upstart-powered core personal loans originated in the fourth quarter of 2023 through the third quarter of 2025 is currently expected to deliver annual returns in line with a blended target of approximately 11.3% after servicing fees. At a more granular level, all quarterly vintages of core personal loans originated in the fourth quarter of 2023 and the first quarter of 2024 are currently forecasted to underperform relative to their target returns. Even though our underwriting models have over time utilized more variables and data points about borrowers, which has improved model performance, they were not designed to predict the severe impact changing macroeconomic conditions, credit market volatility and interest rate fluctuations that occurred following the COVID-19 pandemic, all of which were (and still are) beyond our control. The forecasted underperformance for these vintages reflects the impact of a combination of factors that occurred during that period, including the elimination of government stimulus measures and the worsening of the macroeconomic environment, via rising inflation and the resulting sharply higher interest rates.

The core personal loans originated in the second quarter of 2024 or later are currently forecasted to deliver returns in line with target yields. This reversion in performance reflects a combination of factors including increased conservatism in underwriting, the relative stabilization of macroeconomic conditions, and improvements in our more recent models.

Impact of Macroeconomic Environment

In addition to impacting credit performance, the macroeconomic environment has a direct and indirect impact on our business financial condition, and results of operation. In an economic downturn, we believe consumer lending will generally contract. Lending partners and institutional investors will generally require higher rates of return, which in turn increases the interest rates offered to borrowers, leading to lower borrower demand. Macroeconomic factors can also cause fluctuations of available capital in our lending marketplace due to shifts in the risk preferences of our lending partners and institutional investors. We expect these dynamics would generally invert in an economic upswing.

For example, loan funding provided by institutional investors started to become constrained in 2022, largely due to concerns about the macroeconomic environment. Rising interest rates also led to more expensive loan offers across borrower categories, which decreased borrower demand. In order to create greater stability for our business, we began securing committed capital and co-investment arrangements with institutional investors and other third parties that provide loan funding over longer durations. While we believe that the macroeconomic environment started to improve in 2024, disruption in financial markets could once again lower borrower demand or impair our lending partners and result in constrained funding, which would adversely impact our business, financial condition and operating results.

To respond to macroeconomic changes and provide relevant and up-to-date information to our lending partners, we introduced a new metric, the Upstart Macro Index (“UMI”), in 2023. As of December 31, 2025, UMI

57

Table of Contents

Upstart Holdings, Inc.

Management’s Discussion and Analysis of Financial Condition and Results of Operations

(Tabular Amounts in Thousands, Except Share and Per Share Data and Ratios, or as Noted)

remained elevated, measured at approximately 1.39, meaning that current macroeconomic conditions contributed an incremental risk of approximately 39% to the repayment performance of an Upstart-powered unsecured personal loan, compared to the baseline measurement of 1.0.

Key Operating and Non-GAAP Financial Metrics

We focus on several key operating and Non-GAAP financial metrics to measure the performance of our business and help determine strategic direction. The following presents our key operating and financial metrics:

_______

(1)Transaction Volume, Number of Loans, is shown in ones for the years presented.

(2)Beginning in the fourth quarter of 2025, we revised the definition and underlying calculation methodology of Conversion Rate. Prior period figures have been recast to conform to the new definition and methodology. See discussion of “Conversion Rate” below for further information.

(3)Beginning in the fourth quarter of 2025, we revised the definition and underlying calculation methodology of Percentage of Loans Fully Automated. Prior periods have not been adjusted, as the impact was immaterial. See discussion of “Percentage of Loans Fully Automated” below for further information.

(4)Represents a non-GAAP financial measure. See the section titled “Management’s Discussion and Analysis of Financial Condition and Results of Operations—Reconciliation of Non-GAAP Financial Measures” for further information.

Transaction Volume

We define Transaction Volume, Dollars as the total principal of loan originations (or committed amounts for HELOCs) facilitated on our marketplace during the periods presented. We define Transaction Volume, Number of Loans as the number of loan originations (or commitments issued for HELOCs) facilitated on our marketplace during the periods presented. We believe these metrics are good proxies for our overall scale and reach as a marketplace.

Transaction Volume is driven by improvements in our AI models and technology, including our ability to streamline and automate the loan application and origination process. Transaction Volume can also be driven by several other factors, including borrower acceptance rates and their sensitivity to the interest rates offered through our platform. Transaction Volume is dependent on the availability of platform funding which is influenced by factors such as volatility in the capital markets and macroeconomic conditions.

Transaction Volume, Dollars increased 86% in the year ended December 31, 2025 compared to the prior year and Transaction Volume, Number of Loans increased 115% in the year ended December 31, 2025 compared to the prior year. These increases were primarily due to model improvements and product initiatives, which resulted in an increase in the number of qualified borrowers and more attractive loan offers. The increase in Transaction Volume, Number of Loans was higher than the increase in Transaction Volume, Dollars due to the decrease in

58

Table of Contents

Upstart Holdings, Inc.

Management’s Discussion and Analysis of Financial Condition and Results of Operations

(Tabular Amounts in Thousands, Except Share and Per Share Data and Ratios, or as Noted)

average loan size, primarily as underwriting model improvements drove higher approval rates in smaller dollar categories of loans.

Conversion Rate

We define Conversion Rate as the Transaction Volume, Number of Loans in a period divided by the total number of rate inquiries received that we estimate to be legitimate, which we record when a borrower actively requests a loan offer on our platform. We track this metric to understand the impact of improvements to the efficiency of our borrower funnel on our overall growth. Beginning in the fourth quarter of 2025, we made two adjustments to revise the definition and underlying calculation methodology of Conversion Rate to better align with how management evaluates the efficiency of our borrower funnel and to better support our multi-product business. First, under the new methodology, an application that does not qualify for an offer for one product but is automatically priced for a different product would count as one rate inquiry during that period rather than two rate inquiries under the old methodology. Second, multiple unfunded applications submitted by a single borrower for the same product across multiple quarters are now considered separate inquiries in each quarter under the new methodology, rather than being consolidated as one inquiry under the old methodology. These changes in aggregate resulted in a net decrease in historical Conversion Rates by 0 to 3 percentage points during 2023 and 2024. Prior period Conversion Rate metrics have been recast to conform to the revised definition and methodology to ensure comparability across periods.

Historically, our Conversion Rate has benefited from improvements to our technology, which have made our evaluation of risk more accurate and our verification process more automated, or from the addition of capital partners that have made our offers more competitive. However, our Conversion Rate can be impacted by a variety of internal factors such as changes in the amount of platform and referral fees that we charge or changes in the rate of returns we target for our lending partners and institutional investors. External factors such as shifts in macroeconomic conditions, including interest rate changes, also impact our Conversion Rate. Our ability to continue to improve our Conversion Rate depends in part on our ability to continue to improve our AI models and Percentage of Loans Fully Automated and the mix of marketing channels in any given period.

Our Conversion Rate increased to 19.4% in the year ended December 31, 2025 from 15.1% in the year ended December 31, 2024, primarily driven by underwriting model improvements and product and pricing initiatives, coupled with continued optimization in our acquisition channels.

Percentage of Loans Fully Automated

A driver of our Contribution Margin and operating efficiency is the Percentage of Loans Fully Automated, which is defined as the total number of loans in a given period originated end-to-end with no human involvement required by the Company divided by the Transaction Volume, Number of Loans in the same period. Under this definition, “originated end-to-end” means (i) from initial rate request to final funding for personal loans, including small dollar loans, and (ii) from initial rat

[Excerpt truncated for page length; the complete text is on the linked full-MD&A page.]