Executive Summary / Key Takeaways

• Upstart Holdings, Inc. is rapidly solidifying its leadership in AI-driven lending, leveraging proprietary models and extensive data to transform consumer credit with superior risk assessment and automation.
• The company achieved GAAP profitability in Q2 2025, a quarter ahead of schedule, and sustained strong financial performance in Q3 2025, with GAAP net income growing sixfold over the prior quarter, demonstrating significant operating leverage.
• Strategic expansion into Auto Lending and Home Equity Lines of Credit (HELOCs) is gaining significant traction, with these newer products showing triple-digit year-on-year growth and contributing meaningfully to new borrower acquisition.
• Upstart's commitment to "best rates, best process for all" is driving competitive advantages, particularly in the super prime segment, supported by continuous model innovations like Model 22 and embeddings.
• While macroeconomic conditions and regulatory scrutiny present ongoing risks, Upstart's adaptive AI models and diversified funding strategy are designed for resilience, positioning the company for continued growth and market share expansion in 2026 and beyond.

The AI Revolution in Lending: Upstart's Foundational Approach

Upstart Holdings, Inc. is at the forefront of a profound transformation in the consumer credit industry, aiming to radically reduce the cost and complexity of borrowing for all Americans through its proprietary artificial intelligence (AI) lending marketplace. Founded in 2012, Upstart's mission is to remake the entire lending process by applying advanced AI models and cloud applications to underwriting. This differentiated approach, which the company refers to as "risk separation," has generally led to higher borrower approvals and lower interest rates compared to traditional lending practices, while also providing more predictable returns for its capital partners.

The company's overarching strategy is to become the "always-on, everything-store for credit," persistently underwriting 100% of Americans with the best credit products available. This vision is underpinned by a continuous cycle of AI model refinement and expansion into new product categories. Upstart's AI models extend beyond core underwriting to encompass income and identity verification, loan stacking, and servicing, driving significant levels of automation. For instance, during the nine months ended September 30, 2025, an impressive 91% of loans on its platform were fully automated, requiring no human intervention from Upstart.

The competitive landscape in consumer lending is dynamic and highly competitive, with a mix of traditional banks, non-bank lenders, and other fintech platforms. Upstart competes with diversified digital banks like SoFi (SOFI), peer-to-peer lenders such as LendingClub (LC), buy-now-pay-later providers like Affirm (AFRM), and specialized mortgage lenders such as Rocket Companies (RKT). Upstart distinguishes itself through its pure-play AI focus, emphasizing technological innovation in credit modeling rather than a broad marketplace or full-service financial ecosystem. This specialized approach allows Upstart to potentially offer more accurate risk assessment and faster processing speeds, which are critical differentiators in attracting both borrowers and lending partners.

Technological Edge: The Engine of Growth

Upstart's core competitive advantage lies in its sophisticated AI technology, which is continuously upgraded, trained, and refined, drawing on over ten years of development. This creates powerful "flywheel effects," enhancing the sophistication of its algorithms and the volume of training data. Unlike other AI platforms, Upstart generates its own training data, boasting more than 98 million borrower repayment events to date, with approximately 105,000 new repayment events added daily. This vast dataset is crucial for driving improved risk separation and model accuracy.

Recent technological advancements underscore this leadership. Model 18 introduced the innovation of using the loan's Annual Percentage Rate (APR) as an input, or "feature," significantly enhancing model accuracy. In Q4 2024, Model 19 launched the Payment Transition Model (PTM), which considers intermediate delinquency states in training data, allowing the model to learn from loans that recover from delinquency. This leads to more accurate models by leveraging previously underutilized data. Further, Model 22, launched in early May 2025, integrated neural networks at every level of its architecture, increasing Upstart's separation accuracy advantage over benchmark credit models by 17 percentage points to 171.2%. This effectively decreased the remaining inaccuracy to be solved by 87.5%.

Operational benefits from these technological advancements are quantifiable and impactful. Latency in loan pricing has been reduced by as much as 30% through parallelization, enabling larger and more complex models. In direct marketing channels, a proprietary technique leveraging causal impact to conversion has shown a 50% uplift in incremental originations for the same level of spend. For newer products, multimodal AI is being deployed to automate human document review for HELOCs in real-time, and instant property valuations have driven automatic home loan approvals from less than 1% in June 2025 to about 20% in October 2025. The auto refinance product saw a remarkable 7x improvement in conversion rates across 2024, and the first instant approval of an auto refinance loan was completed in just 9 minutes in Q1 2025. These innovations are not merely incremental; they are foundational shifts that enhance efficiency, reduce costs, and improve the borrower experience across all product lines.

Financial Performance and Strategic Momentum

Upstart's financial performance in 2025 reflects a strong return to growth and profitability, driven by its technological advancements and strategic initiatives. For the nine months ended September 30, 2025, total revenue reached $747.77 million, with revenue from fees, net, at $684.79 million, representing a 57% year-on-year increase. The company achieved GAAP net income of $34.97 million for the nine months ended September 30, 2025, and a notable $31.81 million in Q3 2025 alone, marking a sixfold increase over the prior quarter and achieving GAAP profitability a quarter earlier than anticipated. Adjusted EBITDA for the nine months ended September 30, 2025, was $166.79 million, with an Adjusted EBITDA margin of 22%.

Transaction volume in dollars increased 80% year-on-year in Q3 2025 to $2.85 billion, while the number of loans grew 128% year-on-year to 428,056. This divergence is partly due to underwriting model improvements driving higher approval rates in smaller dollar loan categories and a strategic mix shift. The conversion rate, a key metric for efficiency, increased to 20.60% in Q3 2025 from 16.30% in Q3 2024, reflecting model improvements and acquisition channel optimization. However, a temporary model conservatism in Q3 2025, in response to an uptick in the Upstart Macro Index (UMI), caused a sequential dip in conversion rates from Q2 2025's 23.9%. Management views this as a "speed bump," with the UMI since reverting to lower numbers and no material deterioration in consumer credit strength.

Operating expenses demonstrate a disciplined approach. While sales and marketing expenses increased by 82% year-on-year in Q3 2025 due to higher advertising and borrower acquisition costs, customer operations expenses decreased as a percentage of total revenue from 24% to 18% in Q3 2025, reflecting efficiencies from automation. Engineering and product development expenses also decreased as a percentage of total revenue from 40% to 23% in Q3 2025, despite ongoing investments in AI models. The company's contribution margin in Q3 2025 was 57%, reflecting a slight sequential decrease due to lower conversion rates but remaining robust.

Upstart's liquidity position is supported by cash and cash equivalents of $489.78 million as of September 30, 2025. The company held $1.23 billion in loans on its balance sheet, with 71% allocated to R&D for new products like auto and HELOCs. Upstart plans to reduce these R&D-related holdings in Q4 2025 and into 2026, which is expected to moderate net interest income from this source.

The company maintains a diversified funding strategy, with over 50% of loan funding from committed capital and co-investment arrangements. It has 10 active institutional funding partners and recently issued a securitization in September 2025 with strong demand, involving 30 investors.

Expanding Horizons and Competitive Dynamics

Upstart's strategic expansion into Auto Lending and Home Equity Lines of Credit (HELOCs) is proving to be a significant growth driver. These newer products accounted for almost 12% of originations and 22% of new borrowers in Q3 2025, with transaction volume for each growing approximately 300% year-on-year.

`

The auto retail business, in particular, has seen accelerating momentum, doubling live lending rooftops and growing transaction volume over 70% sequentially in Q3 2025. The company is also developing an "auto secured personal loan" hybrid product. For HELOCs, Upstart offers "best rates to the primest borrowers" compared to other fintechs, by as much as 300 basis points, and has achieved automatic home loan approvals for about 20% of applications in October 2025.

In the competitive landscape, Upstart's AI-driven approach provides a distinct advantage. While competitors like SoFi offer diversified financial services and LendingClub operates a broad marketplace, Upstart's deep specialization in AI for credit assessment allows it to achieve superior risk separation and automation. This technological edge enables Upstart to offer competitive rates and a frictionless process, which is crucial in attracting and retaining borrowers. The company's ability to generate its own training data from over 98 million repayment events is a unique moat, making it difficult for competitors to replicate its model accuracy.

Upstart's funding strategy, which includes a mix of bank and credit union partners, institutional investors, and securitization programs, provides resilience against market fluctuations. While some competitors may have larger balance sheets or more diversified revenue streams, Upstart's 100% retention rate with its private credit partners and strong demand for its securitizations demonstrate confidence in its AI platform's ability to deliver consistent returns. The company is actively securing third-party funding for its newer products, with 17 partner agreements signed in 2025, including 9 in Q3 alone, to transition funding off its balance sheet.

However, risks persist. The Upstart Macro Index (UMI), while improving, remains elevated at approximately 1.54 as of September 30, 2025, indicating ongoing macroeconomic risk. The credit performance of core personal loans originated from Q3 2023 through Q1 2024 is still forecasted to underperform, though later vintages are on target. Regulatory scrutiny of AI in financial services is increasing, with new "true lender" laws and AI regulations emerging at state and federal levels, which could impact Upstart's business model. The company's reliance on its balance sheet for R&D loans also exposes it to credit risk. Furthermore, the rising trend in repayment speeds, while positive for consumer health, can temporarily limit interest income and necessitate higher coupons.

Outlook and Conclusion

Upstart's outlook for Q4 2025 and the full year 2025 reflects continued growth and a commitment to profitability. For Q4 2025, the company expects total revenues of approximately $288 million, with revenue from fees at $262 million and net interest income at $26 million. GAAP net income is projected to be $17 million, and Adjusted EBITDA is expected to reach $63 million. For the full year 2025, Upstart anticipates total revenues of approximately $1.035 billion, with revenue from fees of $946 million and net interest income of $89 million. The Adjusted EBITDA margin is guided at approximately 22%, and GAAP net income is expected to be positive for the full calendar year.

These projections are based on conservative macroeconomic assumptions, including a stable UMI and no explicit expectation of Fed rate cuts. Management plans to strategically moderate take rates to invest in customer lifetime value, aiming for higher origination volumes and repeat transactions. The company is "all systems go" for a strong finish to 2025 and anticipates an "amazing 2026," driven by continued model improvements, the scaling of new products, and the expansion of its funding network.

Upstart's journey from a nascent AI startup to a GAAP-profitable, rapidly growing fintech leader underscores the transformative power of its technology. The company's relentless pursuit of AI innovation, evidenced by its continuously improving models and expanding product suite, positions it uniquely in the multi-trillion-dollar credit market. While macroeconomic headwinds and evolving regulatory landscapes present challenges, Upstart's disciplined execution, robust funding strategy, and unwavering focus on delivering "best rates, best process for all" create a compelling investment thesis. The company's ability to generate its own training data and translate AI breakthroughs into tangible financial and operational benefits suggests a strong foundation for sustained market share gains and long-term value creation.